Cardiology Journal Cardiology Journal ClubClub
Sanjay Dravid, M.D.Sanjay Dravid, M.D.
January 17, 2006January 17, 2006
MULTIPLE BIOMARKERS FOR MULTIPLE BIOMARKERS FOR THE PREDICTION OF FIRST THE PREDICTION OF FIRST MAJOR CARDIOVASCULAR MAJOR CARDIOVASCULAR
EVENTS AND DEATH EVENTS AND DEATH
Wang, Thomas J., et al. Wang, Thomas J., et al. Massachusetts General Hospital. Massachusetts General Hospital. NEJM. Volume 355(25), 21 December 2006, NEJM. Volume 355(25), 21 December 2006,
pp 2631-2639.pp 2631-2639.
OverviewOverview
To evaluate the incremental usefulness of To evaluate the incremental usefulness of multiple biomarkers from various pathways.multiple biomarkers from various pathways.
Established risk factors, including smoking, htn, Established risk factors, including smoking, htn, DM, and dyslipidemia.DM, and dyslipidemia.
Significant interest in new biomarkers for risk Significant interest in new biomarkers for risk stratification of ambulatory persons. stratification of ambulatory persons.
Novel ApproachNovel Approach
Many individual biomarkers have been studied.Many individual biomarkers have been studied. ““Multimarker” ApproachMultimarker” Approach Simultaneous measurement may enhance risk Simultaneous measurement may enhance risk
stratification?stratification?
Outcomes AnalysisOutcomes Analysis
1. Death from any cause1. Death from any cause 2. 12. 1stst Major cardiovascular event (MI, coronary Major cardiovascular event (MI, coronary
insufficiency, heart failure, and stroke.insufficiency, heart failure, and stroke. Reviewed by a committee of three investigators Reviewed by a committee of three investigators
Study SampleStudy Sample
Large, community based cohort studyLarge, community based cohort study Participants from the sixth examination cycle Participants from the sixth examination cycle
(1995-1998) of the Framingham Offspring Study (1995-1998) of the Framingham Offspring Study IRB of Boston University Medical Center IRB of Boston University Medical Center
approval approval Written informed consent was obtained Written informed consent was obtained H & P, PE, and Lab AssessmentH & P, PE, and Lab Assessment
Exclusion CriteriaExclusion Criteria
Serum creatinine levels greater than 2.0 mg/dLSerum creatinine levels greater than 2.0 mg/dL Missing covariatesMissing covariates Prior event when determining outcome of major Prior event when determining outcome of major
cardiovascular eventcardiovascular event Triglycerides > 400Triglycerides > 400
Biomarker Selection Biomarker Selection
1. Marker of inflammation- hsCRP1. Marker of inflammation- hsCRP 2. Markers of neurohormonal activity- BNP, 2. Markers of neurohormonal activity- BNP,
aldosterone, renin, N-terminal pro-atrial aldosterone, renin, N-terminal pro-atrial natriuretic peptidenatriuretic peptide
3. Marker of thrombosis and inflammation- 3. Marker of thrombosis and inflammation- fibrinogenfibrinogen
4. Marker of fibrinolytic potential and 4. Marker of fibrinolytic potential and endothelial function- plasminogen-activator endothelial function- plasminogen-activator
Biomarker cont’dBiomarker cont’d
Inhibitor type 1Inhibitor type 1 5. Marker of thrombosis- D-dimer5. Marker of thrombosis- D-dimer 6. Marker of endotheial function and oxidant 6. Marker of endotheial function and oxidant
stress- homocysteinestress- homocysteine 7. Marker of glomerular endothelial function- 7. Marker of glomerular endothelial function-
urinary albumin-to-creatinine ratio urinary albumin-to-creatinine ratio
Lab ProtocolLab Protocol
Fasting blood and urine samples collected in Fasting blood and urine samples collected in morning after patient supine for ~10 minutes. morning after patient supine for ~10 minutes. Immediately centrifuged and stored at -70 Immediately centrifuged and stored at -70 degreesC. degreesC.
Standardized Assay MethodsStandardized Assay Methods
Statistical AnalysisStatistical Analysis
Multivariable proportional-hazards model (2 sets Multivariable proportional-hazards model (2 sets of analyses for each outcome due to urine of analyses for each outcome due to urine subgroups)subgroups)
Logarithmic transformation used to normalize Logarithmic transformation used to normalize the distribution of biomarkersthe distribution of biomarkers
To reduce the number of false positives from To reduce the number of false positives from multiple testing: multiple testing:
Statistics cont’dStatistics cont’d
1) Multivariable Cox regression model1) Multivariable Cox regression model 2) Backward elimination 2) Backward elimination 3) Construction of multimarker score3) Construction of multimarker score 4) Quintiles categorized 4) Quintiles categorized 5) Cumulative probability curves constructed by 5) Cumulative probability curves constructed by
the Kaplan-Meier method for low, intermediate the Kaplan-Meier method for low, intermediate and high mulitmarker scoresand high mulitmarker scores
Statistics cont’dStatistics cont’d
Then calculated hazard ratios for death and Then calculated hazard ratios for death and major cardiovascular events for the mulitmarker major cardiovascular events for the mulitmarker score groups score groups
Adjusted for age, sex, conventional risk factors Adjusted for age, sex, conventional risk factors including htn, smoking, dm, etc.including htn, smoking, dm, etc.
““C statistic”C statistic” ROC curvesROC curves
Statistics cont’dStatistics cont’d
Secondary Analysis adjusting for medication useSecondary Analysis adjusting for medication use Repeated a Cox proportional-hazards model for Repeated a Cox proportional-hazards model for
major cardiovascular events adjusting for major cardiovascular events adjusting for “nonmajor events” “nonmajor events” angina, intermittent angina, intermittent claudication, TIAclaudication, TIA
SAS software, version 8 (SAS Institute) SAS software, version 8 (SAS Institute)
C StatisticC Statistic
Defined as the probability of concordanc among Defined as the probability of concordanc among persons who can be compared.persons who can be compared.
Estimated as the sum of concordance values Estimated as the sum of concordance values divided by the number of comparable pairs.divided by the number of comparable pairs.
Better able to measure discrimination than Better able to measure discrimination than relative risk. relative risk.
ResultsResults
Total of 3532 persons- 21 excluded for serum Total of 3532 persons- 21 excluded for serum creatinine and 302 for missing covariates.creatinine and 302 for missing covariates.
10 year follow-up (median 7.4 years) 3209 10 year follow-up (median 7.4 years) 3209 available for study.available for study.
207 (6%) died, of whom 72 were women207 (6%) died, of whom 72 were women 169 (6%, excluding prevalent CV disease at 169 (6%, excluding prevalent CV disease at
baseline) had a major cardiovascular event, of baseline) had a major cardiovascular event, of whom 68 were womenwhom 68 were women
Results cont’d Results cont’d
Biomarker panel for nine: P<0.001 for death Biomarker panel for nine: P<0.001 for death and P=0.005 for cardiovascular eventsand P=0.005 for cardiovascular events
Biomarker panel for ten (2750 persons): Biomarker panel for ten (2750 persons): P<0.001 for death and P=0.04 for P<0.001 for death and P=0.04 for cardiovascular events cardiovascular events
Results cont’dResults cont’d
Backward elimination models: final statistical Backward elimination models: final statistical model included only the following biomarkers:model included only the following biomarkers:
BNP, homocysteine, urinary albumin-to-creatinine BNP, homocysteine, urinary albumin-to-creatinine ratio and renin for death.ratio and renin for death.
BNP and urinary albumin-to-creatinine ratio for BNP and urinary albumin-to-creatinine ratio for major cardiovascular events.major cardiovascular events.
Utility of Multimarker ScoresUtility of Multimarker Scores
Backward elimination biomarkers selected as Backward elimination biomarkers selected as statistically significant were incorporated into statistically significant were incorporated into mulitmarker scores.mulitmarker scores.
Restricted to urine sample patients: 1) death Restricted to urine sample patients: 1) death from any cause, the number of events and from any cause, the number of events and number at risk were 172 and 2750, respectively; number at risk were 172 and 2750, respectively; 2) major cardiovascular events, 133 and 2598, 2) major cardiovascular events, 133 and 2598, respectively.respectively.
Utility?Utility?
Persons with high multimarker scores had a risk Persons with high multimarker scores had a risk of death four times as great and a risk of major of death four times as great and a risk of major cariovascular events almost two times as great as cariovascular events almost two times as great as persons with low mulitmarker scores.persons with low mulitmarker scores.
(P<0.001 and P=0.02, respectively) (P<0.001 and P=0.02, respectively)
DiscussionDiscussion
~10 year study of biomarkers indicating BNP, ~10 year study of biomarkers indicating BNP, hsCRP, homocysteine, renin, and alb/Cr ratio as hsCRP, homocysteine, renin, and alb/Cr ratio as most informative for predicting death, while most informative for predicting death, while BNP and alb/Cr ration as significant for BNP and alb/Cr ration as significant for predicting cardiovascular outcome.predicting cardiovascular outcome.
Although high multimarker scores conferred Although high multimarker scores conferred greater risk for death and major cardiovascular greater risk for death and major cardiovascular events…events…
ConclusionConclusion
Mulitmarker scores (combination of biomarkers) Mulitmarker scores (combination of biomarkers) add only moderately to conventional risk factors add only moderately to conventional risk factors as evidenced by small changes in C statistic.as evidenced by small changes in C statistic.
Single biomarkers may have correlation with Single biomarkers may have correlation with predicting outcomespredicting outcomes
Panel likely will not be useful or cost-effective in Panel likely will not be useful or cost-effective in ambulatory setting for further risk stratification ambulatory setting for further risk stratification
LimitationsLimitations
Biomarker selection: omission of lipoprotein-Biomarker selection: omission of lipoprotein-associated phospholipase A2associated phospholipase A2
Each individual marker not independently testedEach individual marker not independently tested Not a true cohort study to asses for primary Not a true cohort study to asses for primary
prevention as “nonmajor” cardiovascular events prevention as “nonmajor” cardiovascular events adjusted adjusted
Adiposity or insulin resistance not taken into Adiposity or insulin resistance not taken into account account
SummarySummary
Biomarkers from multiple, biologically distinct Biomarkers from multiple, biologically distinct pathways are associated with the risks of death pathways are associated with the risks of death and major cardiovascular events. and major cardiovascular events.
However, only moderately adds to conventional However, only moderately adds to conventional risk factors currently.risk factors currently.