Red Cell Distribution Width (RDW) as a Novel Prognostic Marker in Heart Failure: Data from the CHARM Program and the Duke Databank
Study Objective n To identify prognostic biomarkers from a wide range of routine laboratory measures in symptomatic heart failure patients Derivation Dataset: CHARM Replication Dataset: Duke Databank
Derivation Dataset n CHARM North American Cohort l N=2679 l Enrollment March March 2001 l Median follow-up of 34 months l Major enrollment criteria included: Symptomatic chronic HF > 4 weeks duration Serum Cr < 3 mg/dl, K < 5.5 mmol/L Absence of MI or stroke in the prior 4 weeks Absence of non-cardiac disease judged to limit 2- year survival
Methods n n Blood samples were collected at time of randomization l n=1085 CHARM-Preserved (EF > 40%) l n=931 CHARM-Added (concurrent ACEI + EF ≤ 40%) l n=663 CHARM-Alternative (ACEI intolerance + EF ≤ 40%) n n 36 measures: l l Chemistry: BMP, Ca/Mg/P, LFTs, lipids, CK, uric acid l l Hematology: CBC, differential, HbA1c n n CV death and HF hospitalizations were the 1º endpoint n n Cox proportional hazards modeling
Clinical Variables by HF Hosp or CV Death Baseline CharacteristicsNo event Event (N=1727) (N=952) (N=1727) (N=952) History Age (year) Etiology (ischemic)65.0%71.5% NYHA Class III56.1%70.1% NYHA Class IV1.3%5.5% Prior hospitalization for HF62.4%77.7% Ejection Fraction Stroke9.4%12.7% Diabetes Mellitus30.8%48.8% Atrial fibrillation25.3%35.2% Physical Exam Systolic BP (mmHg) Pulmonary crackles10.5%16.8% Cardiomegaly8.4%20.1%
Key Laboratory Parameters by Outcome Mean SD Sodium (mmol/l) Urea nitrogen (mg/dl) Creatinine (mg/dl) Albumin (g/dl) Bilirubin Total (mg/dl) Uric Acid (mg/dl) Cholesterol (mg/dl) White cell count (10 3 /mm 3 ) Lymphocytes (%) Hemoglobin (g/dl) RDW (%) Glycohemoglobin A1C (%) CV Death or HF Hospitalization Event (N=952) Event (N=952) No event (N=1727) No event (N=1727)
Multivariable Model for CV Death or HF Hosp CHARM – LaboratoriesHR*95%CI Χ 2 RDW Bilirubin Total Lymphocyte % Uric Acid HbA1C Hemoglobin Creatinine *HR for continuous variables shown as standardized HR (HR per 1 SD)
Multivariable Model for CV Death or HF Hosp CHARM – All VariablesHR*95%CI Χ 2 Age (per 10yr >60) Cardiomegaly RDW HF Hosp < 6 mo Bilirubin Total NYHA Class IV NYHA Class III EF (per 5% ↓<45%) *HR for continuous variables shown as standardized HR (HR per 1 SD)
Red Cell Distribution Width? n RDW = the variation in red blood cell volume (anisocytosis) n Elevated in variety of diseases l Iron deficiency l Malnutrition l Chronic kidney disease n Calculated automatically on every CBC
CHARM Adjusted HR by quintile of RDW for CV Death or HF Hospitalization Adjusted Hazard Ratio RDW Quintiles Events = 952
CHARM Adjusted HR by quintile of RDW for All-Cause Mortality Adjusted Hazard Ratio RDW Quintiles Events = 625
Replication Dataset – Duke Databank n Duke Databank for Cardiovascular Disease (DDCD) l Registry of all patients undergoing cardiac cath at Duke since 1969 l Mortality follow up >96% n Limited dataset for this analysis l l Symptomatic HF (NYHA II-IV) at enrollment Irrespective of LVEF l RDW value available 0-30 days prior to enrollment n Multivariable Cox model was constructed identifying the relationship of baseline variables (including RDW) to all-cause mortality
Multivariable Model for All-Cause Death Duke - All Predictors HR*95%CI Χ 2 Age RDW Hemoglobin # Diseased Vessels Non-CV Charlson Index # Systolic Blood Pressure Ejection Fraction History of Hypertension Male *Hazard ratios for continuous variables shown as standardized hazard rations (HR per 1 SD); #Charlson index is a combined measure of non-cardiac comorbidity
Adjusted Hazard Ratio RDW Quintiles Duke Adjusted HR by quintile of RDW for All-Cause Mortality Events = 368
Conclusions n Among 36 routine laboratory measures in a large HF trial database, higher RDW showed the greatest association with CV death and HF hospitalization n This finding was replicated in a large registry, where RDW continued to be strongly associated with mortality n Understanding how and why this marker is associated with outcome may provide l Improved targeting of HF therapies l Increased understanding HF pathophysiology