Population versus Personalized Medicine in the Clinical Management of CV Disease Discussion Regarding Individualized Approaches to the Management of CV Risk Robert M. Honigberg, MD Key Questions Do management strategies and treatment goals developed from population based studies lead to appropriate guidelines and care of individual patients? Where is the greatest opportunity for individualization related to cardiovascular risk risk assessment or LDL management? 1
Should we treat the line or the dot? 4000 3500 y = 9.7334x + 428.75 R² = 0.5627 3000 LDL-P (nmol/l) 2500 2000 1500 1000 500 0 0 50 100 150 200 250 300 LDL-C (mg/dl) Compared current JNC-7 guidelines with individualized guidelines for blood pressure management to calculate risk reduction from expected from treatment Used ARIC study to perform person-specific analysis with longitudinal data Individualized management could prevent the same number of MIs and strokes as JNC-7 at a savings of 67% or Individualized management could prevent 43% more heart attacks and strokes for the same cost as treatment according to JNC-7 guidelines Ann Intern Med. 2011;154:627-634 2
Agenda Residual cardiovascular risk NCEP ATP Guidelines Risk assessment LDL management Review of key literature MESA, Framingham LDL variability Diabetes and metabolic disease Apo B variability Therapy Health economic model Incidence of CHD and Cholesterol Framingham Heart Study Indicates that Measuring Cholesterol Does Not Tell Us Enough The two distribution curves for CHD and no CHD have 80% overlap of total cholesterol (TC) (150-300 mg/dl) 35% of CHD occurs in people with TC<200 Twice as many individuals with TC<200 had CHD than those with TC>300 Castelli et al. Atherosclerosis 124 Supp (1996) S1-S9 From Framingham Heart Study, at 26 years follow-up 3
40 Residual Cardiovascular Risk in Major Statin Trials CHD events occur in patients treated with statins Patients Experiencing Major CHD Events, % 30 20 10 28.0 19.4 15.9 12.3 13.2 10.2 11.8 8.7 0 4S 1 LIPID 2 CARE 3 HPS 4 WOSCOPS 5 AFCAPS/ TexCAPS 6 N 4444 9014 4159 20 536 6595 6605 LDL -35% -25% -28% -29% -26% -25% Secondary High Risk Primary Placebo Statin 10.9 7.9 5.5 6.8 1 4S Group. Lancet. 1994;344:1383-1389. 2 LIPID Study Group. N Engl J Med. 1998;339:1349-1357. 3 Sacks FM, et al. N Engl J Med. 1996;335:1001-1009. 4 HPS Collaborative Group. Lancet. 2002;360:7-22. 5 Shepherd J, et al. N Engl J Med. 1995;333:1301-1307. 6 Downs JR, et al. JAMA. 1998;279:1615-1622. Residual CVD Risk in Patients Treated With Intensive Statin Therapy Patients Experiencing Major CVD Events, % 8 40 Statistically significant, but clinically inadequate CVD reduction 1 30 20 10 0 N LDL-C,* mg/dl 26.3 22.4 Standard statin therapy Intensive high-dose statin therapy PROVE IT-TIMI 22 2 IDEAL 3 TNT 4 4162 8888 10 001 95 62 104 81 101 77 *Mean or median LDL-C after treatment 13.7 12.0 10.9 8.7 1 Superko HR. Br J Cardiol. 2006;13:131-136. 2 Cannon CP, et al. N Engl J Med. 2004;350:1495-1504. 3 Pedersen TR, et al. JAMA. 2005;294:2437-2445. 4 LaRosa JC, et al. N Engl J Med. 2005;352:1425-1435. 4
Registry Evidence of Residual CVD Risk Among 136,905 hospitalizations for acute CAD events; lipids w/in 24 hrs of admit (at 541 hospitals) Over 50% of CHD patients had LDL-C <100 and 17.6% had LDL-C <70 For patients without h/o CHD, 72.1% had LDL-C <130 and 41.5% had LDL<100 Sachdeva A, et al. Am Heart J 2009; 157:111-7.e2. From AHA s Get with The Guidelines (GWTG) CAD Program and Database; 2000-2006 NCEP ATP III: Determining LDL-C Goals Presence of CHD, DM 2 major CV risk factors* Yes No Yes No 10-year CHD risk: FRS Yes No High-Risk: <100mg/dL, optional <70mg/dL >20% 10-20% <10% High-Risk: <100mg/dL Mod-high Risk: <130mg/dL, optional <100mg/dL Moderate risk <130mg/dL Lower risk <160mg/dL 5
Improve Risk Assessment Opportunities for Improvement? Multivariable Risk Assessment High Risk CHD/CHD Risk Equiv. (>20% 10-year risk)** NCEP ATP III Approach Treatment Assessment & Goal LDL Management Lifestyle/statin/other If above goal LDL-C < 100 Improve LDL Management Intermediate Risk 2+ Risk Factors* (10-20% 10-year risk)** Treatment Assessment & Goal LDL-C < 130 Low Risk 0-1 Risk Factor* (0-10% 10 year risk)** Treatment Assessment & Goal LDL-C < 160 *Risk factors are age (>45M;>55W), smoking, hypertension, low HDL-C, family history **10-year risk given by age, gender, TC, HDL-C, SBP/hypertension, and smoking (Framingham Risk Score) Issues to be addressed by ATP IV CVD risk assessment More stringent targets versus a fixed dose strategy adjusting dose to risk hs-crp Alternative treatment targets: Role of advanced lipoprotein testing Apo B, LDL-P, non-hdl-c Direct targeting of HDL-C and triglycerides Role of fibrates, niacin, ezetimibe Role of imaging of subclinical atherosclerosis 6
NHLBI Integrated Cardiovascular Risk Reduction Guidelines Release of integrated guidelines for: Cholesterol Guideline Update (ATP IV) Hypertension Guideline Update (JNC 8) Obesity Guideline Update (Obesity 2) Also developing new guidelines for: Risk assessment Lifestyle Three major questions in ATP IV What evidence supports LDL-C goals for primary prevention? For secondary prevention? For optimal statin management? What evidence supports LDL-C goals for primary prevention? 7
For more information or to check status: http://www.nhlbi.nih.gov/guidelines/indevelop.htm Improve Risk Assessment Opportunities for Improvement? Multivariable Risk Assessment High Risk CHD/CHD Risk Equiv. (>20% 10-year risk)** NCEP ATP III Approach Treatment Assessment & Goal LDL Management Lifestyle/statin/other If above goal LDL-C < 100 Improve LDL Management Intermediate Risk 2+ Risk Factors* (10-20% 10-year risk)** Treatment Assessment & Goal LDL-C < 130 Low Risk 0-1 Risk Factor* (0-10% 10 year risk)** Treatment Assessment & Goal LDL-C < 160 *Risk factors are age (>45M;>55W), smoking, hypertension, low HDL-C, family history **10-year risk given by age, gender, TC, HDL-C, SBP/hypertension, and smoking (Framingham Risk Score) 8
Recommendations from the NLA Expert Panel on Clinical Utility of Inflammatory Markers and Advanced Lipoprotein Testing Measurement for On-Treatment Management Decisions Davidson, et al. J Clin Lipidol 2011;5:338-367. Measures of LDL are not always equal LDL-P measure different from LDL-C LDL-P measures the number of particles LDL-C measures cholesterol content Apo B POLAR SURFACE COAT Phospholipid Free cholesterol NONPOLAR LIPID CORE Cholesteryl Ester Triglyceride 9
LDL C can vary with particle size At the same LDL cholesterol, more small LDL vs. large LDL particles present Up to 70% More Particles 100 mg/dl 100 mg/dl Large LDL Small LDL Cholesterol Balance Otvos JD et al. Am J Cardiol 2002;90(suppl):22i-29i Cromwell WC et al. J Clin Lipidology. 2007;1(6):583-592. Insulin resistance & LDL-C variability Cholesterol content varies with disease & metabolic state Normal Cholesterol Content Lower Cholesterol Content 100 mg/dl 100 mg/dl Cholesterol Triglycerides Clin Cardiol 1999; 22(6 Suppl):1121-1127 Cholesterol Balance TG enriched particle Cholesterol (ester) exchanged for TG via CETP 10
Atherogenic Dyslipidemia of Insulin Resistance Framingham Offspring Study LDL Particles (nmol/l) 1800 1600 1400 1200 1000 LDL Particles LDL Cholesterol 180 1800 160 1600 140 1400 120 1200 100 1000 LDL Particles LDL Cholesterol 180 160 140 120 100 LDL Cholesterol (mg/dl) 20 40 60 80 100 0 100 200 300 400 HDL Cholesterol (mg/dl) Triglycerides (mg/dl) Cromwell WC and Otvos JD. Curr Athero Reports 2004;6:381-387 High LDL Particle Number Drives Atherogenic Plaque Formation A gradient driven process, LDL particles invade the arterial wall and set in motion the cascade of events that leads to atherosclerosis 1,2 1. Fredrickson et al. NEJM 1967; 276: 148 2. Brunzell, et al. Diabetes Care. 2008;4:811-822 3. Ip et al. Ann Intern Med. 2009;150:474-484 After adjustment for LDL-P concentration, particle subclass and size don t impact outcomes 3 11
Recommendations from the NLA Expert Panel on Clinical Utility of Inflammatory Markers and Advanced Lipoprotein Testing Measurement for On-Treatment Management Decisions Davidson, et al. J Clin Lipidol 2011;5:338-367. Comparative Literature is Population-based* Study CHD Status Atherosclerotic Endpoint LDL-P Associations Stronger? Women s Health Study Circulation 2009; 119:931-9- Primary Prevention Incident MI, CHD death, CVA YES VA-HIT Circulation 2006;113:1556-63 Secondary Prevention Non- fatal MI or CHD Death YES MESA Atherosclerosis 2007;192:211-17. Primary Prevention Carotid IMT; CVD Event YES Framingham Heart Study J Clin Lipidology 2007;1:583-92. Primary Prevention Incident CVD Events YES EPIC-Norfolk Atherosclerosis 2007;49:547-53. Primary Prevention Incident CAD Events YES Cardiovascular Health Study ATVB 2002; 22:1175-1180 Primary Prevention Incident MI or Angina YES PLAC-I Am J Cardiol 2002;90:89-94. Secondary Prevention Angiographic MLD YES Healthy Women Study Am J Cardiol 2002;90(suppl):71-77i. Primary Prevention EBCT Coronary Calcium Score YES *Over 330 NMR-based publications in the peer reviewed literature 12
Principles for determining appropriate LDL testing guided by landmark publication Value of the new reference test is best examined in cases of disagreement (discordance) between the new and standard tests. Clinical consequences of disagreements between new and standard test requires a fair umpire test. Fair umpire tests include clinical events or disease progression. Glasziou P et al. Ann Intern Med 2008;149:816-22. Multi-Ethnic Study of Atherosclerosis (MESA) Large NHLBI observational study of the pathogenesis and progression of subclinical atherosclerosis Subjects include 6,814 asymptomatic men and women free of cardiovascular disease at entry Population consists of 38% White, 28% African-American, 22% Hispanic, and 12% Asian (of Chinese descent) Outcomes in Current Study Carotid IMT of participants not exposed to lipid medications (n=4499) 319 CVD events (MI, CHD death, angina, stroke, stroke death, or other atherosclerotic or CVD death) during 5.5-yr follow-up (n=5598) Otvos et al. J Clin Lipidol 2011;5:105-113 13
Discordance Between LDL-C and LDL-P LDL-P Consistently Tracks Better with Incident CV Events in Discordant Patients Cumulative CV Incidence 0.06 0.04 0.02 LDL-P > LDL-C Concordant LDL-P < LDL-C LDL-C underestimates LDL-attributable risk LDL-C overestimates LDL-attributable risk LDL-C 104 117 130 mg/dl LDL-P 1372 1249 1117 nmol/l 12.5 events/1000 person years 24% increase 10.1 events/1000 person years 27% decrease 7.3 events/1000 Person years 0 1 2 3 4 5 Follow-up (years) Otvos et al. J Clin Lipidol 2011;5:105-113 14
MESA: Characteristics by LDL Discordance Subgroups Bad Discordance LDL-P > LDL-C Concordant LDL-P LDL-C Good Discordance LDL-P < LDL-C Diabetes (%) 17 a 11 7 a Waist (cm) 99 a 96 93 a SBP (mmhg) 127 b 126 125 Glucose (mg/dl) 109 a 104 100 a HOMA-insulin resistance 2.2 a 1.7 1.4 a TG (mg/dl) 161 a 123 99 a HDL-C (mg/dl) 44 a 51 57 a Values are adjusted for age, sex, and race; a p<0.0001; b p<0.01; c p<0.05 vs concordant subgroup Framingham Offspring Study Long-running NIH/NHLBI observational study of residents of Framingham MA to determine risk factors for future CVD Blood samples were obtained in 1988-91 (exam 4). Lipoprotein particles were measured by NMR spectroscopy and lipids by traditional chemical methods CVD occurrence was monitored during 15-year followup There were 431 CVD events (MI, stroke, CHD death, angina, congestive heart failure) among 3,066 subjects Freedman DS et al., Clin Chem 2004;50:1189-1200 Kathiresan S et al., Circulation 2006;113:20-29 Cromwell W et al., J Clin Lipidol 2007;1:57-64 15
DF4 CHD Event Associations of LDL-P vs LDL-C Framingham Offspring Study Cromwell WC et al. J Clin Lipidol 2007;1(6):583-592 Framingham Offspring CVD Event Rates Differences between quartiles within LDL-P 16
Slide 31 DF4 eliminate white "Discordant" lines in graph and from the legend Elimate build David Frey, 5/9/2012
Example in actual payer population LDL-P distribution at LDL-C<100 mg/dl, N= 1586 33% Concordant 67% Discordant 600 522 551 500 Number of Subjects 400 300 200 100 327 162 24 0 LDL-P Concentrations (nmol/l) <1000 1000-1299 1300-1599 1600-2000 >2000 Coronary Heart Disease ICD-9: 410.00-410.92, 411.1-411.89,412, 413.0-413.9, 414.0,414.07, 414.2, 414.8, 414.9, 429.2, 429.9 Diabetes Mellitus ICD-9: 250.00-250.93 Hyperlipidemia ICD-9: 272.0-272.4, 272.8-272.9 Peripheral Arterial Disease ICD-9: 440.0-440.9, 443.81, 443.89 Symptomatic Carotid Artery Disease ICD-9: 433.00-436, 437.0-437.1, 438.0-438.9 LDL-P Discordance in Metabolic Syndrome Metabolic syndrome patients in Framingham Study (n=1138) N=286 N=407 N=355 N=233 N=113 N=30 LDL-C (mg/dl) 180 170 160 150 140 130 120 110 LDL-C LDL-P 1 2 3 4 5 6 0 1 2 3 4 5 MetSyn (-) MetSyn (+) 1800 1700 1600 1500 1400 1300 1200 1100 LDL-P (nmol/l) From Framingham Offspring Study Circulation 2006;113:20-29 17
LDL Particle Number Distribution in T2DM Subjects Percent of Subjects 20 15 10 5 1% (n=19) 5 th 20 th 50 th 80 th percentile 24% (n=364) 43% (n=631) 21% (n=307) 25% 75% 11% (n=163) LDL-C 70-99 mg/dl (n=1,484) 0 20 15 16% (n=147) 700 1000 1300 1600 (nmol/l) 43% (n=377) 30% (n=260) 9% (n=76) 2% (n=15) Percent of Subjects 10 40% LDL-C < 70 mg/dl 5 (n=871) 0 700 1000 1300 1600 (nmol/l) Cromwell WC, Otvos JD. Am J Cardiol 2006;98:1599-1602 2008 ADA/ACC Consensus Statement ApoB or LDL particle number also appear to be more discriminating measures of the adequacy of LDL lowering therapy than are LDL cholesterol or non-hdl cholesterol. *ADA has included LDL particle measurement in 2013 Standards of Medical Care 18
2013 AACE Algorithms for Diabetes LDL-P Distribution in T2DM Subjects, LDL-C<50 mg/dl 8% Concordant (LDL-P<500 nmol/l and Non-HDL-C<80 mg/dl) (% of Subjects) <2nd 8% (n=95) 20% (n=250) (N= 1,242) 5th 20th 50th 80th 47% 21% 4% (n=588) (n=258) (n=43) 25% 0.1% (n=8) 500 Malave et al. Am J Cardiol 2012 700 1000 1300 1600 (nmol/l) 19
Clinical Implications for LDL Management Improve Risk Assessment Multivariable Risk Assessment High Risk CHD/CHD Risk Equiv. (>20% 10-year risk)** Current ATPIII Approach Treatment Assessment & Goal LDL Management Lifestyle/statin/other If above goal LDL-C < 100 Improve LDL Management 56.5% discordance* Reclassify Risk Using New/Different Variables Intermediate Risk 2+ Risk Factors* (10-20% 10-year risk)** Low Risk 0-1 Risk Factor* (0-10% 10 year risk)** Treatment Assessment & Goal Treatment Assessment & Goal LDL-C < 130 LDL-C < 160 LDL-P > >1000 45.3% discordance* LDL-P > >1300 *from MESA *Risk factors are age (>45M;>55W), smoking, hypertension, low HDL-C, family history **10-year risk given by age, gender, TC, HDL-C, SBP/hypertension, and smoking (Framingham Risk Score) To assess comparability of apo B and LDL-P in association with clinical outcomes 85 clinical outcomes in 25 clinical studies Only apo B statistically significant, LDL-P not significant Only LDL-P statistically significant, apo B not significant Both statistically significant Neither statistically significant 5 13 50 17 The level of statistical significance, as indicated by the P value, and the strength of association, as indicated by the OR, RR, and HR, was more often higher for LDL-P than it was for apo B Clinical Chemistry February 2013;59:1-19 20
Discordance Rate Between LDL-P and apo B? The discordance rate was 29.2% In most cases, LDL-P was a stronger predictor than apo B Discordance rates between LDL-P and apob in various subgroups: - Prediction of CVD or events = 21.1% - Presence of carotid atherosclerosis = 29% - Association with diabetes mellitus = 12.5% Clinical Chemistry 2013;59:1-19 Strengths with the NMR Method The NMR measurement of LDL-P appears to be generally more precise than that of apo B Between-run precision, apo B CV s ranging from 5% to 11% Whereas the same measure for LDL-P was from 2% to 4% NMR LipoProfile is a well-standardized, reagentless test, monitored by FDA central lab & regional devices Apo B is reagent-based immunoassay, with WHO criteria for standardization but no way to enforce the standards NMR provides more information with the analysis than does the single measurement of apo B LDL-P by NMR is more specific for LDL-related risk and does not include other apo B associated lipoproteins with shorter half-lives Clinical Chemistry 2013;59:1-19 21
How can LDL-P help manage heart disease? Guide therapy intervention to reduce the number of cardiovascular events Fine tuning medical therapy Avoiding costly hospitalizations and procedures Self-insured employers becoming interested in using LDL-P within disease management programs Can more directly address medication non-adherence LDL-P can be more useful in identifying metabolic patients that are non-adherent Treatments Change LDL-C and LDL-P Differentially Cholesterol per particle decreases with: Statins Statin + Ezetimibe or Bile Acid Sequestrants Estrogen Replacement Therapy Anti-retrovirals (some) Low fat, High carb diet Therapy LDL-C More Than LDL-P Cholesterol per particle increases with: Fibrates Niacin Pioglitazone Omega 3 FAs Exercise Mediterranean and low carb diet Therapy LDL-P More Than LDL-C Cromwell WC. In: Clinical Challenges in Lipid Disorders.Toth PP, Sica DA, editors. Oxford: Clinical Publishing; 2008.p. 249-259. 22
Adapted from Rosenson et al. Atherosclerosis. 2010; 213:1-7. Appropriate Patient Selection Criteria for LDL-P Clinical use of LDL-P is appropriate within the following patient populations being managed with a lipid-lowering therapy or intensive lifestyle management with one of the following criteria: Known cardiovascular disease (CVD); or CHD risk equivalents such as Type II Diabetes Mellitus, Chronic Kidney Disease (CKD); or Multiple cardio-metabolic risk factors that define the Metabolic Syndrome High Blood pressure or on anti-hypertensive medication High blood sugar High triglycerides Low HDL-C Abdominal obesity 23
Medical Policy and Reimbursement for CPT 83704 Medicare -covers CPT 83704 Medicaid - covers CPT 83704 (can vary per state) Tricare - covers CPT 83704 WellPoint/Anthem - retired policy, reimburses CPT 83704 United Healthcare - retired policy, reimburses CPT 83704 Coventry - retired policy, reimburses CPT 83704 BCBS of Alabama covers CPT 83704 with ICD-9 restrictions BCBS of North Carolina - reimburses CPT 83704 CareFirst Blue Cross - reimburses CPT 83704 Magna Care - reimburses CPT 83704 BCBS of Florida- retired policy, reimburses CPT 83704 Regence Blue - retired policy, reimburses CPT 83704 Premera - retired policy, reimburses CPT 83704 Harvard Pilgrim - reimburses CPT 83704 (for diabetes) BCBS of Louisiana - retired policy, reimburses CPT 83704 BCBS of Delaware - retired policy, reimburses CPT 83704 Health economic model of LDL management Adaptation from health economic model under review 24
Health Economic Model for LDL Management Purpose: To assess the cost-effectiveness and health system effects of management to different LDL goals LDL-C & LDL-P combination vs. LDL-C alone Publication also compares LDL-P alone to LDL-C alone Baseline Model has a 3-year time horizon Results reported for years 1 through 3 Based on a hypothetical cohort of 1,000 or 100,000 patients All inputs are based on published literature and referenced in the interactive model Model represents both the payer and the self-insured employer perspectives Cost-Effectiveness Analysis Economic analysis comparing relative costs and outcomes of two or more strategies Formula: Costs LDL-C & LDL-P Costs LDLC Effects LDL-C Effects LD-C & LDL-P Incremental cost effectiveness ratio (ICER) The cost to avoid one additional CVD event with new strategy compared to old Less than $50,000 = cost effective $50,000-$100,000 = border line More than $100,000 = not cost effective 25
Health System Effects: Budget Impact Analysis Cardiovascular events Difference in the number of CVD events between 2 strategies Reduction in heart attack and stroke Costs LDL-C & LDL-P Costs LDLC Cost-savings Cost-savings associated with difference in CVD events Costs of heart attack (fatal and non-fatal) and stroke, years 1-3 Costs of testing and additional drug therapy Hypothesis Cost-savings related to avoidance of CVD events outweigh the additional cost of testing, treating (and more people alive and healthy) Additional cost-savings in model for employers From avoiding indirect costs related to productivity Absenteeism, quality of life, and presenteeism while at work Derived from the number of Quality Adjusted Life Years (QALYs) gained Model flow 26
Total covered lives at elevated risk Results Incremental cost-effectiveness ratio (ICER) 27
Results cumulative number of CVD events avoided Results underlying budget impact Costs of combo Budget Item Cost of lipid alone 22.2 mm MI (fatal & nonfatal) 30.6 mm 28.8 mm Stroke 39.5 mm 8.8 mm Angina 12.1 mm 1.3 mm Other CVD 1.7 mm 100.9 mm Testing/Treatment 80.2 mm 162 mm Total cost over 3 years 164.1 mm - saving costs - saving lives 28
Results increased worker productivity and quality of life Concluding Remarks As quality of care and individual patient outcomes become more important in terms of health care and reimbursement, practitioners must have the appropriate data to optimally treat individual patients rather than the population as a whole. This is demonstrated today in the case of LDL-lowering therapy and hypertension. Only through subgroup analyses do these differences become apparent. Individualized approaches to LDL management can be cost-effective and reduce the overall budget impact through the avoidance of cardiovascular events 29
Compared current JNC-7 guidelines with individualized guidelines for blood pressure management Used ARIC study to perform person-specific analysis with longitudinal data Individualized guidelines can prevent the same number of MIs and strokes as JNC-7 at a savings of 67% or Individualized guidelines can prevent 43% more heart attacks and strokes for the same cost as treatment according to JNC-7 guidelines Archimedes will proactively assess JNC-8 and ATP IV Ann Intern Med. 2011;154:627-634 59 Opportunity to individually manage patients with insulin resistance Prevent under-treatment of CVD risk 4000 3500 y = 9.7334x + 428.75 R² = 0.5627 3000 LDL-P (nmol/l) 2500 2000 1500 1000 500 0 0 50 100 150 200 250 300 LDL-C (mg/dl) 30
Q&A 31