Introduction

Warfarin sodium is the mainstay of oral anticoagulation therapy in the US and many other countries. When dosed properly it is a highly efficacious drug for maintaining sufficient, but not excessive, levels of anticoagulation. However, there is large interindividual variability in warfarin dose requirements.¹ In addition, African Americans have been noted to require higher doses of warfarin than Caucasians.^{2, 3}

The above observations have led to investigations aimed at identifying functional genetic variants that alter the response to warfarin. Two such sets of variants within the cytochrome P-450 CYP2C9 enzyme and the vitamin K epoxide reductase complex 1 (VKORC1) have been clearly associated with a lower warfarin maintenance dose requirement.^{4, 5} The genotypes that are associated with these lower dose requirements are relatively common in Caucasians but rare in African Americans.^{2, 3} Thus, one possible explanation for higher dose requirements in African Americans is the relative rarity of these variants. Another possible explanation is the presence of other functional genetic variants in African Americans.

One such possible source of variability in individual and race-specific warfarin responses is the lipoprotein, apolipoprotein E (APOE). Because warfarin affects coagulation by inhibiting the regeneration of reduced vitamin K from oxidized vitamin K in the liver, the availability of hepatic vitamin K is likely to alter the dose requirement of warfarin.⁶ The main circulating form of vitamin K, phylloquinone, is bound to chylomicrons and chylomicron remnants within the plasma, and APOE influences chylomicron remnant uptake by the liver.⁷ There are three common variants of the APOE gene (designated 2, 3 and 4) that encode the three common isoforms of the APOE molecule; their prevalence varies by race, with 4 being more common in African Americans than in Caucasians.⁸ Each isoform varies in its ability to facilitate clearance of vitamin-K-rich lipoproteins from plasma, with 4 having the greatest clearance.^{9, 10, 11, 12} Plasma vitamin K levels also vary by APOE genotype, with the highest levels among 2 allele carriers and the lowest in 4 carriers.^{10, 11, 12} Accordingly, some investigators have proposed that the lower plasma levels of vitamin K in 4 carriers lead to lower anticoagulant requirements,¹³ whereas others have suggested that the greater uptake of vitamin K into the liver leads to increased availability of vitamin K for coagulation proteins and thus a higher anticoagulant drug requirement to achieve adequate anticoagulation in 4 carriers.¹⁴ To date, observational studies of the effect of APOE genotype on anticoagulant requirements have produced contrasting results.^{13, 14, 15, 16}

The INR Adherence and Genetics (IN-RANGE) study is a prospective cohort study with a primary aim of examining the relationship between APOE genotype and warfarin response. The study also sought to determine the effect of APOE polymorphisms in African Americans, a group not yet studied, independent of the effects of CYP2C9 and VKORC1 variants.

Results

Patient population and distribution of genotypes

The study cohort comprised 232 patients (mean age 58.814.9 years) of whom 121 (52.2%) were self-reported Caucasian and 111 (47.8%) were self-reported African American. The clinical characteristics of the cohort by APOE genotype are shown in Table 1 and the distribution of APOE genotypes within the Caucasian and African American subgroups is shown in Table 2. The allele frequencies for 2, 3 and 4 were, respectively, 7.9, 78.5 and 13.6% in Caucasians and 12.6, 66.2 and 21.1% in African Americans. As expected, the APOE 4 variant was more common in African Americans. The allele distribution was in Hardy–Weinberg equilibrium for both groups (P>0.25). Patients with the APOE 4 variant also were more likely to use more medications that potentiate the effect of warfarin (Table 1). The prevalence of APOE variants was the same among those who reached maintenance dose (APOE 2, 3 and 4 variants: 10, 73 and 17%) versus those who did not (APOE 2, 3 and 4 variants: 10, 72 and 18%).

Table 1 - Clinical characteristics by APOE genotype.

Full table

Table 2 - Distribution of APOE genotype by race.

Full table

Effect of APOE genotype on maintenance dose of warfarin

In the entire cohort, 4 carriers had a significantly higher median weekly maintenance dose (44.4 mg) than non-carriers (35.0 mg; P=0.0043). Those with the 2/2 and 2/3 genotypes had the lowest median maintenance dose (31.2 mg), 3/3 homozygotes had an intermediate maintenance dose (35.0 mg) and those with the 4/3 or 4/4 genotypes had the highest maintenance dose (45.0 mg); these differences were statistically significant (P=0.0118). As explained under Materials and methods, the 10 2/4 patients were excluded a priori from the analysis of the three-level variable. These 10 patients had a median weekly maintenance dose of 35.0 mg, the same as that of the 3/3 homozygotes.

When stratified by race, the warfarin weekly maintenance dose was higher among African Americans (median 37.5 mg, mean 42.9 mg) than Caucasians (median 35.0 mg, mean 36.9 mg), a difference that was statistically significant (P=0.018). The effects of APOE genotype on warfarin maintenance dose, stratified by race, are shown in Table 3. In the multivariable model adjusting for confounders, selected as described under Materials and methods, being an 4 carrier was associated with a statistically significant higher dose in African Americans (multivariable P=0.014) but not Caucasians (multivariable P=0.60, multivariable P value for interaction of genotype by race=0.20, Tables 3 and 4). The three-level variable for APOE genotype showed the highest dose among those with 3/4 or 4/4 genotypes, the next highest among 3/3 homozygotes and the lowest among those with 2/3 or 2/2 genotypes in both African Americans and Caucasians (Tables 3 and 4 and Figure 1). However, the association was statistically significant only within the African American subgroup (multivariable P=0.012), not the Caucasian subgroup (multivariable P=0.81; multivariable P value for interaction=0.056). There were only six patients with the 4/4 genotype and they did not clearly require lower doses. However, the one Caucasian patient with 4/4 was a VKORC1 1173C/T heterozygote, a genotype that is associated with low maintenance doses. In addition, one of the highest dose requirements was among one of the five African Americans with the 4/4 genotype, 105 mg/week. The median dose among those with 2/4 was 35.00 mg in both the few African Americans (n=7) and the few Caucasians (n=3) with this genotype, again consistent with the 3/3 group.

Figure 1.

Median warfarin maintenance dose by APOE genotype. The median warfarin maintenance dose by the three-level APOE variable among Caucasians and African Americans is shown. The bars represent the 25th and 75th percentiles.

Full figure and legend (59K)

Table 3 - Maintenance dose of warfarin by APOE genotype, stratified by race.

Full table

Table 4 - Multivariable models for any 4 and for APOE three-level variable using square root transformed warfarin dose.

Full table

APOE genotype was associated with higher maintenance doses among both African Americans and Caucasians who were homozygous wild type for CYP2C9, but this difference was only statistically significant among African Americans (Table 5). Among subjects with a CYP2C9 variant, there were no statistically significant effects of APOE within either race, but there were very few subjects in these subgroups (Table 5).

Table 5 - Maintenance dose of warfarin by APOE genotype by race, stratified by CYP2C9.

Full table

There was no apparent effect of APOE genotype on time to establishing a maintenance dose in the Caucasian or African American subgroups, when either 4 carrier status (P=0.84 for Caucasians, P=0.46 for African Americans) or the three-level APOE genotype variable (P=0.27 for Caucasians, P=0.63 for African Americans) was considered.

Discussion

This study demonstrates that the APOE 4 allele is associated with a significantly higher warfarin maintenance dose requirement than other APOE alleles and that the relationship between different APOE genotype and dose requirement may follow the same pattern as the published relationships between APOE genotype and hepatic chylomicron clearance.^{7, 9, 10, 11, 12} These findings were independent of CYP2C9 polymorphisms, the VKORC1 1173C/T variant, and numerous other factors that can alter warfarin dose requirements. The results were statistically significant among African Americans but not Caucasians; however, the 4 allele was less common among Caucasians, which could have limited our ability to detect an effect in this group.

Our findings could, at least in part, explain the higher warfarin dose requirement among African Americans that has previously been noted by others.² The 4 allele was associated with a higher maintenance dose of warfarin in African Americans, even after adjusting for the different frequencies of CYP2C9 and VKORC1 classes. Although the APOE genotype may have an effect on maintenance dose in Caucasians, the effect was less pronounced and not statistically significant. Regardless of any potential differences in magnitude of effect between African Americans and Caucasians, the fact that the APOE 4 allele is more common and the CYP2C9 and VKORC1 variants are much less common in African Americans than Caucasians suggests that different maintenance dose requirements could be driven by different prevalences of functionally relevant genetic polymorphisms between these two groups.

Hepatic clearance of chylomicrons, and therefore the vitamin K carried by chylomicrons, is at least partly dependent on APOE genotype, with 4 carriers having the most rapid clearance.¹⁰ Thus, one hypothesis is that the APOE 4 variant facilitates a relatively high uptake of vitamin K by the liver and thus enhances the availability of vitamin K for carboxylation of clotting factors.¹⁴ However, a contrasting hypothesis is that the greater hepatic clearance of vitamin K mandated by the APOE 4 variant does not translate into greater availability of vitamin K for the formation of functional coagulation factors and may, in fact, be associated with reduced availability because of greater catabolism of vitamin K.^{13, 17} Our results support the former hypothesis.

Three prior studies have examined the effects of APOE genotype on warfarin dose requirement. Kohnke et al.¹⁴ studied 183 Swedish patients (33.5% with one or two 4 alleles) and found that, among CYP2C9 wild-type homozygotes, those with the 4/4 genotype required significantly higher warfarin maintenance doses than those with one or no 4 alleles. Although our study had too few 4/4 patients to draw meaningful conclusions about this genotype class, our results are consistent with a similar effect of APOE genotype on maintenance dose, extending this finding to African Americans. A follow-up study by Kohnke et al.¹⁵ failed to confirm the effect of APOE genotype on warfarin dose requirement in Italian patients; however, that cohort had only 20 (14%) 4 carriers and no 4/4 homozygotes. Sconce et al.¹⁶ found that the 4 allele was associated with lower doses in a Caucasian population, but, interestingly, 4 carriers had the highest plasma vitamin K concentrations, not the lowest as might be expected.^{10, 11, 12} Clearly, the relationship between APOE genotype and warfarin dose requirements is complex; further study is needed to better understand the interactions between APOE and race, other genotypes, diet and vitamin K levels. In a retrospective study of different anticoagulants, acenocoumarol and phenprocoumon, Dutch Caucasian 4 carriers and homozygotes required a significantly lower mean dose of acenocoumarol, but not of phenprocoumon, relative to 3/3 homozygotes.¹³ Comparisons between our findings and those of this last study are necessarily limited, however, because of the use of different medications and a different definition of maintenance dose that may not represent steady-state requirements. Furthermore, the significant findings were limited to patients taking acenocoumarol and thus could represent the play of chance. With respect to warfarin specifically, our study is consistent with the prior Swedish study.¹⁴

There are several strengths and potential limitations that must be considered in interpreting our results. Our study was prospective and specifically designed to test the hypothesis that APOE genotype would have an effect on warfarin maintenance dose. The prospective cohort design should prevent selection bias, a common concern in retrospective case–control studies. We also were able to control for numerous other factors that are known to alter warfarin response. However, we were only able to adjust for a single polymorphism in VKORC1. Although this polymorphism has been associated with lower warfarin dose in multiple studies and is just as informative as inferred haplotypes in Caucasians,^{5, 18, 19} further assessment of other VKORC1 polymorphisms is needed. Due to our sample size, we also were limited in our ability to test for interactions by race, and interpretation of our findings must be tempered by the sample size and the possibility of erroneous findings. Our findings among 4/4 patients is also not consistent with prior study, but we had very few patients with this genotype. Finally, our results may not generalize to other populations.

In conclusion, our study demonstrates that APOE genotype may affect warfarin maintenance dose requirements, particularly among African Americans, with higher doses needed by 4 allele carriers and lower doses needed by those with the 2/2 and 2/3 genotypes. These findings could at least partly explain the higher dose requirements noted previously in African Americans. Although the 4/4 genotype was rare and not clearly associated with lower doses in this study, our study suggests that APOE genotype may play some role, along with other factors, in determining warfarin response and may explain some of the racial differences in warfarin dose requirements.

Materials and methods

Study population and data collection

From April 2002 to December 2005, patients were prospectively recruited at three anticoagulation clinics: the Hospital of the University of Pennsylvania (HUP) and the Philadelphia Veterans Affairs Medical Center (PVAMC) in Philadelphia, PA, and the Penn State Milton S Hershey Medical Center (HMC) in Hershey, PA. Patients 21 years and older initiating warfarin therapy once daily with a target INR of 2.0–3.0 who presented to one of the clinics were considered eligible for the study. We excluded patients with abnormal INRs before initiating warfarin and those with anti-phospholipid antibody in whom the INR measurement may not be valid.²⁰ This study also excluded nine patients who were not self-reported Caucasian/white or African American/black.

Information on patient demographics, medical history, medication use, warfarin dose, diet and INR was obtained prospectively by trained study interviewers using standardized questionnaires. Genomic DNA was obtained from buccal swabs and was analyzed by facility staff blinded to patient characteristics or outcomes.

Genotyping

DNA was extracted from buccal swab preparations using a method adapted from Richards et al.²¹ Two sets of swabs were taken from each participant and used for validation of methods and quality control.

APOE genotypes (based on single nucleotide polymorphisms (SNPs) rs429358 and rs7412) were determined by the molecular genotyping and diagnostics facility at HUP using a standard method.²² Briefly, polymerase chain reaction (PCR)-restriction fragment length polymorphism (RFLP) assays were performed in 50-l reactions in PCR buffer that contained 5 l of genomic DNA, 200 nM of the forward ((MDX 216) 5'-GCACGGCTGTCCAAGGAGCTGCAGGC-3') and reverse ((MDX 217) 5'-GGCGCTCGCGGATGGCGCTGAG-3')) primers, 200 M dNTPs, 15 mM MgCl₂, 5.0 l 10% dimethylsulfoxide, and 1.25 U Amplitaq gold. Thermocycling conditions consisted of 1 min at 95°C, followed by 45 cycles of 95°C for 30 s, 65°C for 30 s and 72°C for 30 s, with a final extension step of 72°C for 10 min. The 270-bp product was digested with 1.5U HhaI (New England Biolabs, Ipswich, MA, USA) at 37°C for 2 h. The products were separated on 4% agarose/Nusieve gels and visualized by ethidium bromide staining and UV illumination. The 2 allele yielded bands of 91 and 83 bp; the 3 allele yielded bands of 91, 48 and 35 bp; and the 4 allele yielded bands of 72, 48 and 35 bp.

The region containing the CYP2C9^*2 SNP (rs1799853) was PCR amplified in a 25-l volume that contained 5 l of genomic DNA, 200 nM of the forward (5'-GTATTTTGGC CTGAAACCCATA-3') and reverse (5'-GGCCTTGGTTTTT CTCAACTC-3') primers, 200 M dNTPs, NW3 buffer (16 mM (NH₄)₂SO₄, 67.5 mM Tris, 0.01% Tween 20), 2 mM MgCl₂, and 1.5 U Amplitaq (Roche, Indianapolis, IN, USA). The 452-bp product was digested with 1.5 U AvaII (New England Biolabs) at 37°C for 6 h. The products were separated on 3% agarose gels and visualized by ethidium bromide staining and UV illumination. For CYP2C9^*3 (rs1057910), PCR amplifications were performed using 200 M dNTPs, NW3 buffer, 2 mM MgCl₂ and 1.5 U Hotstar Taq Polymerase (Qiagen, Valencia, CA, USA) in a final volume of 25 l. Genomic DNA (5 l) and 0.2 pg of heteroduplex generator were coamplified in the presence of 200 nM of the forward primer (5'-CAGGAAGAGATTGAACGTGTG-3') and the reverse primer (5'-ACAAACTTACCTTGGGAATGAGA-3'). The heteroduplexes and homoduplexes were separated on 12% polyacrylamide (2.6% crosslinking) minigels run at 150 V for 2 h and visualized using ethidium bromide staining and UV illumination. The CYP2C9^*3 heteroduplex generator was validated against the Nsi I RFLP and gave the same results.

VKORC1 1173C/T (rs9934438) polymorphism was chosen because it has been demonstrated in several studies to be associated with lower dose requirements and is just as informative as inferred haplotypes in Caucasians.^{5, 18, 19} The region containing the VKORC1 1173C/T (based on SNP rs9934438) variant was PCR amplified in 25-l reactions in PCR buffer that contained 5 l of genomic DNA, 200 nM forward (5'-AAGATGAAAAGCAGGGCCTAC-3') and reverse (5'-CCGAGAAAGGTGATTTCCAA-3') primers, 200 M each dNTP, 1X PCR buffer (Qiagen), and 1 U HotStar Taq polymerase (Qiagen). Thermocycling conditions consisted of 95°C for 5 min, followed by 45 cycles of 95°C for 40 s, 59°C for 50 s and 72°C for 40 s, with a final extension step of 75°C for 5 min. The 195-bp product was digested with 2 U StyI (New England Biolabs) at 37°C for 6 h; the 1173T allele is cleaved to yield fragments of 125 and 70 bp. The products were separated on 12% polyacrylamide gels and visualized by ethidium bromide staining and UV illumination.

Outcomes

The primary outcome for the study was the maintenance dose of warfarin, determined using the standard definition as the dose that leads to a stable INR over three consecutive visits following initiation of the drug.⁵ In addition, we examined the time required to reach the maintenance dose.

Statistical analysis

The demographic and clinical characteristics of subjects were compared by APOE genotype using ² or exact tests for categorical variables and Wilcoxon rank sum tests for continuous variables. The Hardy–Weinberg equilibrium test was used to assess whether genotype frequencies were in conformity with predictions based on allele frequencies. APOE genotype was categorized in two ways, first based on the presence or absence of any APOE 4 allele (e.g., 4/4, 4/3 and 4/2 versus non-4), the '4 carrier' group. This was based on prior studies demonstrating that people with an 4 allele clear vitamin-K-rich lipoproteins from the circulation more efficiently and have lower plasma vitamin K concentrations than those who do not carry an APOE 4 allele^{10, 11} and a prior study suggesting that the presence of the 4 allele is associated with higher warfarin dose requirements.¹⁴ Second, we categorized APOE genotypes into three groups (the 'three-level APOE' variable), based on data suggesting that plasma vitamin K levels vary by these groups, with levels being highest among those with the 2/2 or 2/3 genotype, intermediate with 3/3, and lowest with 4/4 or 4/3.¹¹ The 2/4 genotype is rare²³ and vitamin K levels have not been studied in this class; therefore, the 10 2/4 patients were excluded a priori from analysis of the three-level variable.^{10, 12}

The distribution for the warfarin maintenance dose was right-skewed, so transformations were performed to achieve approximate normality using the family of Box–Cox transformation²⁴ using a maximum likelihood criterion,²⁵ resulting in a square-root transformation on the maintenance dose. Linear regression analyses based on this transformation were then performed to test for differences in maintenance dose by APOE genotype and other potential confounders. Covariates that were associated with maintenance dose with a bivariate P value <0.2 were included in a multivariable linear regression model. Cox proportional hazards models were used to examine differences in the time to maintenance therapy by APOE genotype. For these analyses, patients who did not reach maintenance therapy were censored at their last follow-up visits.

All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC, USA) and Stata version 9 (StataCorp, College Station, TX, USA). The study was approved by the Institutional Review Boards at all participating hospitals, and all patients provided informed, written consent.

References

1. Breckenridge AM. Interindividual differences in the response to oral anticoagulants. Drugs 1977; 14: 367–375. | Article | PubMed | ChemPort |
2. Gage BF, Eby C, Milligan PE, Banet GA, Duncan JR, McLeod HL. Use of pharmacogenetics and clinical factors to predict the maintenance dose of warfarin. Thromb Haemost 2004; 91: 87–94. | PubMed | ISI | ChemPort |
3. Takahashi H, Wilkinson GR, Nutescu EA, Morita T, Ritchie MD, Scordo MG et al. Different contributions of polymorphisms in VKORC1 and CYP2C9 to intra- and inter-population differences in maintenance dose of warfarin in Japanese, Caucasians and African-Americans. Pharmacogenet Genomics 2006; 16: 101–110. | PubMed | ChemPort |
4. Sanderson S, Emery J, Higgins J. CYP2C9 gene variants, drug dose, and bleeding risk in warfarin-treated patients: a HuGEnet systematic review and meta-analysis. Genet Med 2005; 7: 97–104. | Article | PubMed | ISI | ChemPort |
5. Rieder MJ, Reiner AP, Gage BF, Nickerson DA, Eby CS, McLeod HL et al. Effect of VKORC1 haplotypes on transcriptional regulation and warfarin dose. N Engl J Med 2005; 352: 2285–2293. | Article | PubMed | ISI | ChemPort |
6. Majerus PW, Broze Jr GJ, Miletich JP, Tollefsen DM. Anticoagulant, thrombolytic, and antiplatelet drugs. In: Hardman JG, Limbird LE, Molinoff PB, Ruddon RW, Goodman Gilman A (eds). Goodman & Gilman's The Pharmacological Basis of Therapeutics. McGraw-Hill: New York, 1996, pp. 1341–1359.
7. Lamon-Fava S, Sadowski JA, Davidson KW, O'Brien ME, McNamara JR, Schaefer EJ. Plasma lipoproteins as carriers of phylloquinone (vitamin K1) in humans. Am J Clin Nutr 1998; 67: 1226–1231. | PubMed | ISI | ChemPort |
8. Corbo RM, Scacchi R. Apolipoprotein E (APOE) allele distribution in the world. Is APOE^*4 a 'thrifty' allele? Ann Hum Genet 1999; 63: 301–310. | Article | PubMed | ISI | ChemPort |
9. Weintraub MS, Eisenberg S, Breslow JL. Dietary fat clearance in normal subjects is regulated by genetic variation in apolipoprotein E. J Clin Invest 1987; 80: 1571–1577. | PubMed | ChemPort |
10. Kohlmeier M, Salomon A, Saupe J, Shearer MJ. Transport of vitamin K to bone in humans. J Nutr 1996; 126: 1192S–1196S. | PubMed | ChemPort |
11. Saupe J, Shearer MJ, Kohlmeier M. Phylloquinone transport and its influence on gamma-carboxyglutamate residues of osteocalcin in patients on maintenance hemodialysis. Am J Clin Nutr 1993; 58: 204–208. | PubMed | ChemPort |
12. Kohlmeier M, Saupe J, Drossel HJ, Shearer MJ. Variation of phylloquinone (vitamin K1) concentrations in hemodialysis patients. Thromb Haemost 1995; 74: 1252–1254. | PubMed | ChemPort |
13. Visser LE, Trienekens PH, De Smet PA, Vulto AG, Hofman A, van Duijn CM et al. Patients with an ApoE epsilon4 allele require lower doses of coumarin anticoagulants. Pharmacogenet Genom 2005; 15: 69–74. | ChemPort |
14. Kohnke H, Sorlin K, Granath G. Warfarin dose related to apolipoprotein E (APOE) genotype. Eur J Clin Pharmacol 2005; 61: 381–388. | Article | PubMed | ChemPort |
15. Kohnke H, Scordo MG, Pengo V, Padrini R, Wadelius M. Apolipoprotein E (APOE) and warfarin dosing in an Italian population. Eur J Clin Pharmacol 2005; 61: 781–783. | Article | PubMed |
16. Sconce EA, Daly AK, Khan TI, Wynne HA, Kamali F. APOE genotype makes a small contribution to warfarin dose requirements. Pharmacogenet Genomics 2006; 16: 609–611. | PubMed | ChemPort |
17. Kohlmeier M, Saupe A, Saupe J. Anticoagulant response to phenprocoumon is related to apolipoprotein E genotype. Klin Lab 1995; 41: 359–361. | ChemPort |
18. D'Andrea G, D'Ambrosio RL, Di Perna P, Chetta M, Santacroce R, Brancaccio V et al. A polymorphism in the VKORC1 gene is associated with an interindividual variability in the dose-anticoagulant effect of warfarin. Blood 2005; 105: 645–649. | Article | PubMed | ISI | ChemPort |
19. Bodin L, Verstuyft C, Tregouet DA, Robert A, Dubert L, Funck-Brentano C et al. Cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase (VKORC1) genotypes as determinants of acenocoumarol sensitivity. Blood 2005; 106: 135–140. | Article | PubMed | ISI | ChemPort |
20. Moll S, Ortel TL. Monitoring warfarin therapy in patients with lupus anticoagulants. Ann Intern Med 1997; 127: 177–185. | PubMed | ChemPort |
21. Richards B, Skoletsky J, Shuber AP, Balfour R, Stern RC, Dorkin HL et al. Multiplex PCR amplification from the CFTR gene using DNA prepared from buccal brushes/swabs. Hum Mol Genet 1993; 2: 159–163. | Article | PubMed | ISI | ChemPort |
22. Tsukamoto K, Watanabe T, Matsushima T, Kinoshita M, Kato H, Hashimoto Y et al. Determination by PCR-RFLP of apo E genotype in a Japanese population. J Lab Clin Med 1993; 121: 598–602. | PubMed | ISI | ChemPort |
23. Davignon J, Gregg RE, Sing CF. Apolipoprotein E polymorphism and atherosclerosis. Arteriosclerosis 1988; 8: 1–21. | PubMed | ISI | ChemPort |
24. Box GEP, Cox DR. An analysis of transformations. J Royal Stat Soc 1964; B-26: 211–252.
25. Draper NR, Smith H. Applied Regression Analysis. Second ed. John Wiley & Sons Inc.: New York, 1981.

Acknowledgements

We thank Sandy Barile for editorial assistance, Joseph A Gascho, MD, for serving as the site investigator at HMC, Frederick F Samaha, MD, for serving as the site investigator at the PVAMC, and Sarah L Booth, PhD, for her critical insights. We are also indebted to Mitchell Laskin, RPh; Mabel Chin, PharmD; and Francis Herrmann, BS, RPh, for their dedication to our field work. Funded by NIH Grant R01HL066176–04; Drs Kimmel and Whitehead are also funded by NIH P20-RR020741. The funders had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; or preparation, review, or approval of the manuscript. Dr Kimmel has received research funding from GlaxoSmithKline and has served as a consultant to Bayer and GlaxoSmithKline, all unrelated to warfarin. The data will be deposited in PharmGKB (www.pharmgkb.org). Funded by NIH grants R01HL066176 and P20RR020741.