Candidates for chronic warfarin therapy often have co-morbid conditions, such as heart failure, with reduced left ventricular ejection fraction. Previous reports have demonstrated an increased risk of over-anticoagulation due to reduced warfarin dose requirement in patients with decompensated heart failure. However, the influence of left ventricular systolic dysfunction (LVSD), defined as left ventricular ejection fraction <40%, on warfarin response has not been evaluated. Here, we assess the influence of LVSD on warfarin dose, anticoagulation control (percent time in target range), and risk of over-anticoagulation (international normalized ratio >4) and major hemorrhage. Of the 1,354 patients included in this prospective cohort study, 214 patients (16%) had LVSD. Patients with LVSD required 11% lower warfarin dose compared with those without LVSD (p <0.001) using multivariate linear regression analyses. Using multivariate Cox proportional hazards model, patients with LVSD experienced similar levels of anticoagulation control (percent time in target range: 51% vs 53% p = 0.15), risk of over-anticoagulation (international normalized ratio >4; hazard ratio 1.01, 95% confidence interval 0.82 to 1.25; p = 0.91), and risk of major hemorrhage (hazard ratio 1.11; 95% confidence interval 0.70 to 1.74; p = 0.66). Addition of LVSD variable in the model increased the variability explained from 35% to 36% for warfarin dose prediction. In conclusion, our results demonstrate that patients with LVSD require lower doses of warfarin. Whether warfarin dosing algorithms incorporating LVSD in determining initial doses improves outcomes needs to be evaluated.
Warfarin has been the mainstay of oral anticoagulant therapy since the 1950s. Despite extensive experience with its use, patients remain in the therapeutic range only half of the time. Whereas subtherapeutic anticoagulation leads to inadequate protection, supratherapeutic anticoagulation can lead to life-threatening bleeding. This significant variability in dose requirements suggests an unmet need for identifying the clinical and genetic predictors of warfarin dose-response in individual patients. Previous studies have identified gender, age, smoking, co-morbid conditions (e.g., chronic kidney disease [CKD]) and the use of concomitant drugs (e.g., amiodarone) as influential clinical predictors, and variation in cytochrome P4502C9 ( CYP2C9 ; the principal metabolic pathway for warfarin) and vitamin K epoxide reductase complex 1 ( VKORC1 ; the target protein inhibited by warfarin) as important genetic predictors. Although, these efforts have improved the prediction of warfarin dose requirements, a significant proportion (40% to 60%) of dose variation remains unexplained. Previous studies have suggested that patients with decompensated heart failure (HF) require lower dose of warfarin. However, the influence of left ventricular systolic dysfunction (LVSD) on warfarin dose requirement, anticoagulation control, and over-anticoagulation and bleeding risk have not been systematically evaluated. Here, we assess the influence of LVSD on warfarin dose, anticoagulation control as measured by percent time in target range (PTTR), risk of excessive anticoagulation, and risk of major hemorrhage while accounting for the known clinical and genetic predictors of warfarin response.
Methods
The prospective Warfarin Pharmacogenetics Cohort recruited patients ≥20 years initiating warfarin therapy with a target international normalized ratio (INR) of 2 to 3 and managed at an anticoagulation clinic under the approval of the Institutional Review Board at the University of Alabama at Birmingham and at Emory University.
Participants were enrolled at the initiation of warfarin therapy and followed up for up to 2 years or for the duration of therapy if <2 years (e.g., for venous thromboembolism). At enrollment, a detailed history was obtained including self-reported race, education, income, medical insurance, height and weight, blood urea nitrogen, serum creatinine, hemoglobin and hematocrit, indication for warfarin therapy, co-morbid conditions, medications, smoking status, alcohol use, and vitamin K intake (number of servings of foods rich in vitamin K consumed per week) as detailed in previous publications.
In addition to VKORC1 (rs9923231), and CYP2C9 (*2 [rs1799853], *3 [rs1057910], *5 [rs28371686], *6 [rs9332131], and *11 [rs28371685]), we also assessed CYP4F2 (rs2108622), and the CYP2C single-nucleotide polymorphism rs12777823 as reported.
LVSD was defined as left ventricular ejection fraction (LVEF <40%) by echocardiography. LVEF was missing in 9.6% of the patients (143 of 1,497), and therefore, these patients were not included in the analysis. Kidney function was assessed using the estimated glomerular filtration rate calculated using the Modification of Diet in Renal Disease Study equation. Patients were categorized into 3 groups: (1) reference group (normal, stage 1 CKD, or stage 2 CKD), (2) stage 3 CKD, and (3) stage 4 or 5 CKD.
Warfarin dose (mg/day; log transformed to attain normality of residuals) was calculated as the average dose required for maintenance of therapeutic anticoagulation. Proportion of PTTR was estimated for each patient using the Rosendaal linear interpolation method. This method assumes that a linear relation exists between 2 consecutively measured INR values and allows one to allocate a specific INR value to each day for each patient. Time in target range for each patient was assessed by the percentage of interpolated INR values within the target range of 2.0 to 3.0 after attainment of first INR in target range. Over-anticoagulation was defined as INR >4.
Hemorrhagic complications were classified as reported by Schulman et al. Major hemorrhage events included serious, life-threatening, and fatal bleeding episodes. During the follow-up, all major hemorrhagic complications were captured and verified through review of admissions and emergency department visits. The Alabama Center for Health Statistics was queried to verify cause of death for all deceased to ensure inclusion of deaths due to hemorrhagic complications. All complications were adjudicated independently by the Medical Director of the Anticoagulation Clinic. Only adjudicated events were included in the analyses.
ANOVA was used to assess group differences for continuous variables and the chi-square test of independence for categorical variables. The Hardy-Weinberg equilibrium assumption was tested using the chi-square test. Multivariable linear regression analysis was performed to evaluate the association between the LVSD on warfarin dose after accounting for sociodemographic (age, race, body mass index), lifestyle (vitamin K and alcohol intake), clinical factors (co-morbid conditions, e.g., atrial fibrillation and diabetes mellitus), concurrent medications (e.g., amiodarone, statins), and genetic factors ( CYP2C9 , and VKORC1 , CYP4F2 , and rs12777823).
To assess the relation between LVSD with over-anticoagulation (INR >4) and risk of major hemorrhagic events, we used the Cox proportional hazards (PH) model with the counting process format. Robust variance estimation was used to correct for the dependence among multiple events per subject and provide 95% confidence intervals (CIs) for the hazard ratios (HRs) of interest. There were no departures from the PH assumption as assessed by evaluating interactions of the predictors and a function of survival time.
We also assessed the influence of LVSD on risk of hemorrhage after adjustment for bleeding risk as assessed by the HAS-BLED score. The score assigns 1 point each to hypertension, abnormal renal (defined as GFR <30 ml/min) and/or liver function, stroke, bleeding history or predisposition, labile INR (defined as PTTR <60%), old age (>65 years), and drug (antiplatelet agent) or alcohol use. Both the cumulative score and the individual component risk factors were evaluated.
All analyses were performed using backward stepwise regression with the full model with exclusion set at p = 0.2. All analyses were performed using SAS, version 9.3, at a non-directional alpha level of 0.05.
Results
The study population included 1,354 patients who received warfarin for therapeutic anticoagulation and were followed for 1.4 ± 0.9 years ( Table 1 ). Among the study population, 16% of the patients had LVSD, defined as LVEF <40%. Indications for anticoagulant therapy in patients with LVSD more commonly included atrial fibrillation, myocardial infarction, or other cardiac-related outcomes/procedures. Patients with LVSD were more likely to be men, and African-American, with higher prevalence of hyperlipidemia, coronary artery disease, chronic kidney disease, and use of statin, antiplatelet, or amiodarone therapy. Possession of variants known to influence warfarin response did not differ by LVEF categories, and their distribution was observed to be in Hardy-Weinberg equilibrium (all p values >0.15).
| Variables | LVSD | P value | |
|---|---|---|---|
| Absent (n=1140) | Present (n=214) | ||
| Mean ± SD | Mean ± SD | ||
| Follow-up (years) | 1.4 ± 0.9 | 1.4 ± 0.9 | 0.85 |
| Visits/person/month | 2.2 ± 1.6 | 2.3 ± 2.0 | 0.49 |
| Age (years) | 61 ± 16 | 61 ± 16 | 0.79 |
| Hematocrit (%) | 37 ± 6.8 | 39 ± 6.4 | <0.001 |
| Body mass index (kg/m 2 ) | 31 ± 7.9 | 29 ± 6.2 | 0.001 |
| Average vitamin K intake | 2.1 ± 1.8 | 1.8 ± 1.3 | 0.08 |
| N (%) | N (%) | ||
| Female | 595 (52%) | 58 (27%) | <0.001 |
| Race | |||
| European American | 655 (58%) | 106 (50%) | 0.03 |
| African American | 485 (43%) | 108 (51%) | |
| Current alcohol use | 326 (28%) | 67 (31%) | 0.42 |
| Current smoker | 134 (12%) | 31 (15%) | 0.26 |
| Indication for warfarin therapy | |||
| Venous thromboembolism | 521 (46%) | 30 (14%) | <0.001 |
| Stroke / Transient Ischemic Attack | 57 (5.0%) | 16 (7.5%) | 0.14 |
| Atrial Fibrillation | 483 (42%) | 109 (51%) | 0.02 |
| Myocardial infarction | 12 (1.1%) | 10 (4.7%) | <0.001 |
| Peripheral arterial disease | 15 (1.3%) | 1 (0.5%) | 0.29 |
| Other | 51 (4.5%) | 48 (22%) | <0.001 |
| Hypertension | 777 (69%) | 149 (70%) | 0.78 |
| Hyperlipidemia | 551 (49%) | 121 (57%) | 0.03 |
| Diabetes mellitus | 362 (32%) | 75 (35%) | 0.38 |
| Coronary artery disease | 328 (29%) | 108 (51%) | <0.001 |
| Chronic Kidney disease | |||
| Stage ≤2 | 719 (63%) | 110 (52%) | 0.005 |
| Stage 3 | 308 (27%) | 79 (37%) | |
| Stage ≥4 | 108 (10%) | 24 (11%) | |
| Concurrent medications | |||
| Statins ∗ | 617 (54%) | 150 (70%) | <0.001 |
| Anti-platelets † | 689 (61%) | 161 (76%) | <0.001 |
| Amiodarone | 109 (10%) | 41 (19%) | <0.001 |
| Percent patients possessing > 1 minor allele ‡ , § | |||
| CYP2C9 variant ¶ | 238 (25%) | 40 (21%) | 0.37 |
| VKORC1 variant ‖ | 425 (43%) | 77 (41%) | 0.64 |
| CYP4F2 variant # | 356 (38%) | 63 (35%) | 0.45 |
| rs12777823 ** | 336 (36%) | 67 (37%) | 0.69 |
∗ Statins included any of the HMG-COA reductase inhibitors.
† Antiplatelet agents included aspirin, clopidogrel, and dipyridamole as mono or dual therapy.
‡ CYP2C9 , and VKORC1 , CYP4F2 , and rs12777823 were categorized as 0 if no variants and 1 if heterozygous or homozygous for the variant allele. CYP2C9 *2, *3, *5, *6, and *11 were combined together to create a single CYP2C9 variable.
§ Genotyping not completed at the time of analysis and therefore genotype information is not available in 194 patients for ¶ CYP2C9 ; 167 patients for ‖ VKORC1 (rs9923231 “T” allele); 225 patients for # CYP4F2 (rs2108622; “A” allele) and 226 patients for ** rs12777823 (“A” allele).
Patients with LVSD had a lower warfarin dose requirements compared with those without LVSD (4.9 and 5.7 mg/day, respectively, p <0.001). Using backward stepwise linear regression, the variables that were retained in the final model are listed in Table 2 . On multivariate adjustment for the earlier mentioned variables, LVSD was independently and significantly associated with lower warfarin dose requirements. LVSD was associated with 11% lower warfarin dose, compared with those without LVSD ( Table 2 , Figure 1 ). Addition of LVSD variable in the model increased the variability explained from 35% to 36% for warfarin dose prediction.
| Variable | Effect on Warfarin Dose (95% CI) | ||
|---|---|---|---|
| Beta coefficient | % Dose Change | p value | |
| Reference ∗ | 2.24 | ||
| Left Ventricular Systolic Dysfunction | -0.165 | -15 (-21 to -9) | <0.001 |
| Female | -0.136 | -13 (-17 to -8) | <0.001 |
| Alcohol use | 0.066 | 7 (1 to 13) | 0.02 |
| Diabetes mellitus | 0.059 | 6 (1 to 12) | 0.03 |
| Chronic Kidney Disease | -0.069 | -7 (-10 to -3) | <0.001 |
| Statin use | -0.051 | -5 (-10 to 0.1) | 0.05 |
| Amiodarone use | -0.208 | -19 (-25 to -12) | <0.001 |
| CYP2C9 variant | -0.275 | -24 (-28 to -20) | <0.001 |
| VKORC1 variant | -0.311 | -27 (-30 to -23) | <0.001 |
| rs12777823 variant | -0.075 | -7 (-12 to -3) | 0.003 |
| Age (per 10 years) | -0.067 | -6 (-8 to -5) | <0.001 |
| Body Mass Index (every 5 units) | 0.056 | 6 (4 to 8) | <0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
