Context Polymorphisms in genes involved in cholesterol synthesis, absorption,
and transport may affect statin efficacy.
Objective To evaluate systematically whether genetic variation influences response
to pravastatin therapy.
Design, Setting, and Population The DNA of 1536 individuals treated with pravastatin, 40 mg/d, was analyzed
for 148 single-nucleotide polymorphisms (SNPs) within 10 candidate genes related
to lipid metabolism. Variation within these genes was then examined for associations
with changes in lipid levels observed with pravastatin therapy during a 24-week
period.
Main Outcome Measure Changes in lipid levels in response to pravastatin therapy.
Results Two common and tightly linked SNPs (linkage disequilibrium r2 = 0.90; heterozygote prevalence = 6.7% for both) were
significantly associated with reduced efficacy of pravastatin therapy. Both
of these SNPs were in the gene coding for 3-hydroxy-3-methylglutaryl-coenzyme
A (HMG-CoA) reductase, the target enzyme that is inhibited by pravastatin.
For example, compared with individuals homozygous for the major allele of
one of the SNPs, individuals with a single copy of the minor allele had a
22% smaller reduction in total cholesterol (−32.8 vs −42.0 mg/dL
[−0.85 vs −1.09 mmol/L]; P = .001; absolute
difference, 9.2 mg/dL [95% confidence interval {CI}, 3.8-14.6 mg/dL]) and
a 19% smaller reduction in low-density lipoprotein (LDL) cholesterol (−27.7
vs −34.1 mg/dL [−0.72 vs −0.88 mmol/L]; P = .005; absolute difference, 6.4 mg/dL [95% CI, 2.2-10.6 mg/dL]).
The association for total cholesterol reduction persisted even after adjusting
for multiple tests on all 33 SNPs evaluated in the HMG-CoA reductase gene
as well as for all 148 SNPs evaluated was similar in magnitude and direction
among men and women and was present in the ethnically diverse total cohort
as well as in the majority subgroup of white participants. No association
for either SNP was observed for the change in high-density lipoprotein (HDL)
cholesterol (P>.80) and neither was associated with
baseline lipid levels among those actively treated or among those who did
not receive the drug. Among the remaining genes, less robust associations
were found for squalene synthase and change in total cholesterol, apolipoprotein
E and change in LDL cholesterol, and cholesteryl ester transfer protein and
change in HDL cholesterol, although none of these met our conservative criteria
for purely pharmacogenetic effects.
Conclusion Individuals heterozygous for a genetic variant in the HMG-CoA reductase
gene may experience significantly smaller reductions in cholesterol when treated
with pravastatin.
Therapy with 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase
inhibitors (statins) lowers total and low-density lipoprotein (LDL) cholesterol
and has proven to be highly effective for cardiovascular risk reduction. However,
there is wide variation in interindividual response to statin therapy, and
it has been hypothesized that genetic differences may contribute to this variation.
While the implications of this hypothesis are broad in terms of "personalized
medicine" and the use of genetic screening to guide selection of lipid-lowering
therapy, clinical data addressing pharmacogenetic interactions with statins
are limited and have largely focused on the lipid metabolism genes apolipoprotein
E (APOE), apolipoprotein B (APOB), cholesteryl ester transfer protein (CETP),
and the LDL receptor (LDLR).1-4
To explore this issue systematically, we genotyped 148 single-nucleotide
polymorphisms (SNPs) across 10 candidate genes known to affect cholesterol
synthesis, absorption, and transport and statin metabolism and sought to correlate
variation within these genes with the change in total, LDL, and high-density
lipoprotein (HDL) cholesterol observed among 1536 individuals treated with
pravastatin during a 24-week period. We evaluated polymorphisms in the genes
encoding HMG-CoA reductase (the target for statin therapy) as well as in the
genes encoding squalene synthase (another key enzyme in cholesterol biosynthesis
and potential target for cholesterol-lowering therapy), 2 cholesterol transport
adenosine triphosphate–binding cassette proteins, APOE, APOB, CETP, LDLR, and the cytochrome P450 system.5-9 Rare
loss-of-function mutations in each of these genes except squalene synthase
and the cytochromes have been shown to result in profound effects on lipid
levels.10 On the basis of prior evidence for
their central role in controlling lipid levels, these genes were chosen as
candidates for examining the hypothesis that their common genetic variants
influence the degree of lipid level reduction during pravastatin therapy.
The study population was derived from participants in the Pravastatin
Inflammation/CRP Evaluation (PRINCE), a community-based randomized trial and
cohort study evaluating the effects of pravastatin, 40 mg/d, or matching placebo
on lipid and inflammatory biomarkers during a 24-week period.11 PRINCE
participants were enrolled from 1143 sites representing 49 states and the
District of Columbia, with no single site enrolling more than 4 patients.
All PRINCE participants were free of statin use in the 6 months prior to enrollment
and had no contraindication to statin therapy. For the current pharmacogenetic
analysis, we limited our evaluation to PRINCE participants who (1) provided
written informed consent for genomic analysis; (2) were randomly allocated
to receive pravastatin or placebo and successfully completed the full PRINCE
study protocol by providing baseline, 12-week, and 24-week blood samples;
and (3) underwent successful DNA extraction and genotyping as outlined herein.
In total, 1536 PRINCE participants assigned to receive pravastatin fulfilled
these criteria and form the basis for these analyses. Of the 1536 study participants,
1362 (88.7%) were self-identified as white, with 100 (6.5%) self-identified
as black, 44 (2.9%) as Hispanic, 19 (1.2%) as Asian, and 11 (0.7%) as other.
This pharmacogenetic study was approved by Brigham and Women's Hospital's
institutional review board.
SNP Selection and Genotyping
We selected candidate genes based on prior observations that null mutations
in all of them, except FDFT1 and the CYP3A genes, grossly alter lipid levels and cause heritable disease.
Squalene synthase encoded by the FDFT1 gene is a
target for cholesterol reduction therapy; CYP3A genes
were included because of their role in statin metabolism, even though they
contribute only modestly in the case of pravastatin.12 Within
the 10 selected candidate genes (Table 1), we identified the common variation due to SNPs by resequencing
in panels of 32 to 96 cell lines from ethnically diverse individuals (roughly
one third European, one fourth Asian, one fourth black, and one sixth Hispanic
or Native American) and by reviewing the literature. By further exploring
linkage disequilibrium, inferred potential effects of SNPs on biological function,
and inferred haplotypes, we were able to winnow our list of candidate SNPs
to a final set of 148 that we believed would capture the important genetic
diversity in the 10 selected genes in the study population. Single nucleotide
polymorphism 12 of the HMG-CoA reductase gene can be found on chromosome 5
at position 74726928 (Human genome July 2003 UCSC version hg16, based on build
34, National Center for Biotechnology Information) by a match to the sequence
AAAAAAAAATTTTTT[AT]AAATCCTTTATATTA, in which the brackets surround the variable
nucleotide. Single nucleotide polymorphism 29 can be found on chromosome 5
at position 74739571 by a match to the sequence TTTTCCAAACTCTTT[TG]GTCATATCAGCCTAA.
A full listing of the 148 SNPs genotyped in the study is available from the
authors.
Genotyping was performed using mass spectrometry–based methods
as described elsewhere.14 Genotypes for HMG
CoA reductase SNPs 12 and 29 were successfully determined in 1504 (97.9%)
and 1518 (98.8%), respectively, of the 1536 participants in the study sample.
A set of reference markers included with the study samples provided no evidence
for inaccurate genotype determinations.
Statistical Analysis and Correction of
To identify potential associations between genotypes in a candidate
gene and response to pravastatin, we first calculated changes in total, LDL,
and HDL cholesterol from the values obtained at baseline to the mean values
of the 12- and 24-week measurements. These changes followed a normal distribution,
allowing us to perform 1-way analysis of variance with F statistics evaluating
the difference in the change in lipid level for, at most, 3 genotypes for
each SNP. Statistical tests were performed using data only from individuals
having mean changes in lipid levels within 3 SDs of the study mean, a constraint
that excluded no more than 25 individuals for any of the lipid values measured.
Only genotype classes populated by at least 10 individuals were considered.
We determined P values by assessing the null
distribution of the F statistic for no association of genotype with the difference
in lipid levels with 10 000 random permutations of genotypes and lipid
values.15P values
determined from the analytic F distribution were essentially the same. Because
HMG-CoA reductase is the target of statin therapy, squalene synthase is an
alternative target for lipid-lowering therapy, and mutations in the other
candidate genes are known to have profound effects on lipid levels, we considered
each gene to represent a strong, independent prior hypothesis. Accordingly,
in addition to the uncorrected P values estimated
from the permutations, we corrected each P value
for the multiple SNPs tested by treating SNPs within a shared gene as a set
of simultaneously assessed hypotheses. Computationally, we examined the null
distribution of the F statistic by rank from permutations involving the set
of SNPs from one gene at a time, and applied a step-down procedure to determine
corrected P values.15 This
permutation approach to correcting P values reflects
the correlated structure of the multiple hypotheses implicit in the linkage
disequilibrium between SNPs belonging to a single locus.
Beyond requiring corrected P<.05 for significance,
we sought to limit the potential for false-positive findings by imposing 2
additional qualitative criteria on any potentially meaningful associations.
First, to reduce possible confounding by race, we insisted that associations
found in the whole racially mixed cohort had to exist as well in the self-identified
white subpopulation, representing 88.7% of the study participants. Second,
while we acknowledged the risk of being too conservative, we were interested
in identifying the best candidates for genetic influences solely on the change
in lipid levels with statin therapy and focused only on SNPs that were not
associated with baseline lipid levels.
In addition to the single-SNP analysis and to uncover epistatic effects
due to combinations of SNPs within the same gene, we tested for associations
between haplotypes and response to pravastatin using an evolutionary approach.16 Haplotypes were inferred with the program PHASE,17 but only those predicted with high confidence were
used in the analysis. Graphically represented as cladograms, evolutionary
trees were derived from the haplotypes through standard assumptions about
allele frequency, allele age, and recombination18 and
through the principle of parsimony, which demands minimization of the number
of nucleotide changes between adjacent (ie, closely related) haplotypes. Once
the trees were constructed, potential associations of the haplotypes and the
changes in lipid values were tested using our algorithm that evaluates all
of the binary partitions of the tree between pairs of adjacent haplotypes.
All genes were examined with the cladistic analysis except CYP3A4, which was genotyped at only 1 SNP. Treescan software was used
for statistical analyses (version 0.8, University of Vigo, Vigo, Spain).
The mean (SD) age of the pravastatin-treated patients was 64.0 (12.5)
years. A total of 534 (34.8%) were women, 211 (13.7%) were current smokers,
and 301 (20%) had diabetes. The mean (SD) baseline levels of total cholesterol,
LDL cholesterol, and HDL cholesterol were 218 (39) mg/dL (5.65 [1.01] mmol/L),
132 (3) mg/dL (3.42 [0.08] mmol/L), and 38 (11) mg/dL (0.98 [0.28] mmol/L),
respectively. These values are consistent with the PRINCE study population
as a whole.
Polymorphism in the HMG-CoA Reductase Gene
Of the 148 SNPs evaluated across the 10 candidate genes, we found 2
tightly linked SNPs that were significantly associated with a difference in
the change in lipid response to pravastatin and that fulfilled our additional
qualitative criteria for association. Both of these SNPs were in the HMG-CoA
reductase gene (SNP 12 and SNP 29) encoding the target for statin therapy,
and the extent of their linkage disequilibrium (r2 = 0.90; P<.001) ensured that the results
for the 2 SNPs were essentially equivalent.
In the study population, genotypes for the 2 SNPs were in Hardy-Weinberg
equilibrium and were discordant for only 7 individuals. As shown in Table 2 for the case of SNP 12, which is
represented as a heterozygous genotype in 6.7% of the study individuals, there
was a marginally significant difference in age by genotype but no significant
differences with regard to sex, traditional risk factors, or pretreatment
lipid levels. Only one participant was homozygous for the minor allele of
either SNP, which was not a sufficient number for inclusion in the statistical
analysis or in the tables.
For individuals with a single copy of SNP 12 in the HMG-CoA reductase
gene, the mean change in total cholesterol associated with pravastatin use
was −32.8 mg/dL (−0.85 mmol/L) while the mean change for those
homozygous for the major allele was −42.0 mg/dL (−1.09 mmol/L),
a reduction in overall efficacy of 21.8% (absolute difference, 9.2 mg/dL [95%
confidence interval {CI}, 3.8-14.6 mg/dL]; P = .001).
For SNP 29 in the HMG-CoA reductase gene, an almost identical 22.3% smaller
effect on total cholesterol reduction was observed (absolute difference, 9.3
mg/dL [95% CI, 3.8-14.7 mg/dL]; P<.001). Both
of these findings remained significant after correction for all 33 SNPs evaluated
in the HMG-CoA reductase gene (both corrected P values
<.02) (Table 3). These effects
were largely due to differences in LDL cholesterol such that individuals heterozygous
for the minor allele experienced an approximate 19% smaller LDL reduction
after taking pravastatin (both P values <.005;
absolute difference for SNP 12, 6.4 mg/dL [95% CI, 2.2-10.6 mg/dL] and absolute
difference for SNP 29, 6.4 mg/dL [95% CI, 2.2-10.7 mg/dL]). In contrast, there
was no significant difference in the change in HDL cholesterol with pravastatin
between genotypes.
We addressed the robustness of these findings in several additional
analyses. First, in the total cohort, the differences in total cholesterol
reduction by genotype persisted even after correction for all 148 SNPs evaluated
across all 10 genes (fully corrected P = .04 for
SNP 12 and P = .049 for SNP 29). To address the potential
for effect modification by sex, we stratified our analysis for all measured
SNPs by sex and, again, found an association for only SNP 12 and SNP 29 in
the HMG-CoA reductase gene. Although the trend of lipid change with genotype
was the same for both sexes, the association was generally more significant
in men.
Second, both SNP 12 and SNP 29 remained significantly correlated with
the reduction of total and LDL cholesterol among whites, representing about
88.7% of the total cohort (Table 3).
The association with the change in LDL cholesterol became stronger after ethnic
stratification; the P value for SNP 29 among whites
was significant after correction for all 33 SNPs in the HMG-CoA reductase
gene.
Third, neither SNP 12 nor SNP 29 was found to contribute to the variance
in baseline levels of total cholesterol or LDL cholesterol, a critical observation
to eliminate the potential for confounding on this basis in detecting purely
pharmacogenetic effects. Not surprisingly, therefore, the 2 SNPs remain statistically
associated with residuals in the change in either total cholesterol or LDL
cholesterol after their regression against baseline lipid level, sex, hormone
therapy status, age, and age squared. Moreover, both of the SNPs in the HMG-CoA
reductase gene remained significantly associated with the residual change
in LDL cholesterol in the subgroup of white participants after correction
for the 33 SNPs in the HMG-CoA reductase gene (P =
.008 and P = .02, respectively).
Finally, we repeated all of these analyses among an additional 649 participants
in the PRINCE trial who did not receive pravastatin therapy. In this group,
we found no evidence of association between the 2 SNPs in the HMG-CoA reductase
gene and lipid levels at baseline or the change in lipid levels after placebo
treatment. These findings markedly reduce the possibility that our primary
observations among those treated represent either regression to the mean or
a nonpharmacological effect of study participation.
Evolutionary analysis of haplotypes inferred from the genotype data
revealed an association entirely equivalent to the genotype-based association
with SNPs 12 and 29. As shown in Figure 1, haplotype 7 at the tip of one branch in the HMG-CoA reductase
cladogram is defined uniquely by the minor alleles of SNPs 12 and 29 and was
the sole haplotype significantly associated with lipid response to pravastatin.
Indeed, the fact that both SNPs 12 and 29 occur uniquely in the same branch
of the cladogram provides a graphical explanation of their high linkage disequilibrium.
In addition to the closely related SNPs in the HMG-CoA reductase gene
already described, 3 other SNPs in the remaining 9 genes were associated with
a differential effect of pravastatin on lipid reduction but failed to meet
at least 1 of our qualitative criteria for association. For example, SNP 4
in the squalene synthase gene (FDFT1) was associated
with the change in total cholesterol in men and persisted in analyses limited
to white men. However, we also found this SNP to be significantly associated
with baseline levels of total cholesterol; thus, after regression against
baseline levels, the association with residuals was no longer significant.
Similarly, while SNP 17 in the APOE gene was
associated with a greater reduction in LDL cholesterol among individuals heterozygous
or homozygous for the minor allele (P = .001; corrected P = .047), it too was associated with baseline LDL levels
(corrected P<.001). Moreover, its effect was attenuated
and no longer fully significant in our analysis limited to white participants
(P = .02; corrected P =
.22). The minor allele of SNP 15 in the cholesteryl ester transfer protein
(CETP) was associated with a smaller increase in
HDL cholesterol among men (P = .007; corrected P = .02) but, again, it was also associated with baseline
HDL cholesterol level (corrected P = .003) and the
effect was not present among women (P = .59; corrected P = .99), among white men (P =
.04; corrected P = .17), or in the sex-combined data
(P = .046; corrected P =
.26). Finally, in the cladistic evolutionary analysis, none of the haplotypes
in genes other than HMG-CoA reductase genes was both significantly associated
with the change in lipid levels in response to pravastatin and not associated
with baseline lipid levels.
In this analysis of 148 SNPs across 10 genes known to be involved in
cholesterol synthesis and statin metabolism, we found 2 common and closely
linked polymorphisms in the HMG-CoA reductase gene that were significantly
associated with a 22% smaller reduction in total cholesterol and a 19% smaller
reduction in LDL cholesterol following 24 weeks of pravastatin therapy. For
total cholesterol, these effects remained significant after adjustment for
all SNPs evaluated and were consistent in magnitude and direction among men
and women and among whites as well as the total cohort.
It is of particular interest that the 2 SNPs with the most robust association
with lipid level reduction after pravastatin therapy both lie within the HMG-CoA
reductase gene, encoding the target for statin therapy. Whether the genetic
effect can be explained by altered expression, activity, or drug binding is
uncertain, but we considered several possible molecular interpretations. The
HMG-CoA reductase gene spans about 24 200 base pairs on chromosome 5q13.3
(July 2003 human genome assembly, National Center for Biotechnology Information
build 34) with an inferred mature transcript of at least 4471 bases spliced
from about 19 exons (RefSeq: NM_000859).19 Single
nucleotide polymorphisms 12 and 29 are separated by about 12 650 base
pairs and reside in introns 5 and 15, respectively, both distant enough from
recognized splicing borders that they are unlikely to interfere with the expected
function of known splicing signals.
In contrast with exons and parts of some introns in the HMG-CoA reductase
gene, sequences surrounding the 2 SNPs are not conserved with sequences from
the mouse genome, suggesting that they may not be directly involved in a conserved
biological process across species. Moreover, neither SNP is part of a CpG
dinucleotide sequence, a potential target for methylation and effects on transcription.
It is thus possible that the SNPs we have identified are linked to other genetic
changes within functional parts of the HMG-CoA reductase gene. For example,
in our multiethnic SNP discovery panel, both SNPs are tightly linked to a
third SNP in a 3′ untranslated exon of the HMG-CoA reductase gene, which
is retained in the mature RNA message. This SNP was not genotyped in the current
study, but it may influence the stability of HMG-CoA reductase RNA in the
cell. We can likely exclude any involvement for SNP 25 encoding the amino
acid substitution I638V in HMG-CoA reductase because this SNP was neither
associated with the change in lipid nor linked to SNPs 12 or 29. Determining
if 1 or more of the HMG-CoA reductase candidate SNPs provides a molecular
explanation of the data observed will thus require further study.
Other than the 2 SNPs in the HMG-CoA reductase gene, we observed no
other SNPs associated with lipid changes following pravastatin therapy using
our conservative analysis plan. The alternative associations that met some
but not all of our qualitative criteria and any others that may exist in our
data may be described adequately only by less conservative statistical constraints
or more complex genetic analysis. For example, we recognize that while our
study had sufficient power to detect large effects due to relatively common
alleles, the power for smaller effects due to rare alleles was limited. Estimated
from the observed variance in the change in total cholesterol among the 1536
participants, the statistical power of our study was sufficient to detect
a change in total cholesterol as small as 9 mg/dL (0.23 mmol/L) for alleles
of 3% or greater frequency. For alleles with a frequency as low as 1%, we
had adequate power to detect a change in total cholesterol of 15 mg/dL (0.39
mmol/L) or greater. Thus, we think it is unlikely that we have missed other
major pharmacogenetic effects.
Our study has limitations that require consideration. First, although
the critical variants observed were in the HMG-CoA reductase gene itself,
we evaluated only pravastatin; thus, care must be taken before generalizing
these data to other statins. Second, the design of our study did not allow
evaluation of dose-response effects. However, the dose of pravastatin used
is the highest dose routinely given for this agent and is the only dose that
has been tested in clinical end-point trials. Third, despite our study's large
sample size and our conservative statistical analysis, these data require
independent confirmation, as would be true for any genetic study. Nonetheless,
it seems unlikely that our study population, which was derived from 49 states
and the District of Columbia, is subject to any major selection bias. Finally,
despite the statistical robustness of our data, the proportion of the variance
that can be explained by HMG-CoA reductase SNPs 12 and 29 is small in comparison
with the expected influence of clinical determinants such as compliance and
diet.
We recognize that these data have considerable pathophysiological interest
and provide strong clinical evidence that there may be promise in the concept
of "personalized medicine" and the use of genetic screening to target certain
therapies. The absolute difference in total cholesterol reduction associated
with the HMG-CoA reductase genotype in our data was 9 mg/dL (0.23 mmol/L),
an effect large enough to affect health on a population basis. Future studies
must determine whether this difference can be offset by dose adjustment or
the choice of an alternative nonstatin lipid-lowering therapy. In the meantime,
clinical reminders to take treatment daily and to titrate dose as necessary
to achieve National Cholesterol Education Program goals remain critical issues
for practice.20
1.Schmitz G, Drobnik W. Pharmacogenomics and pharmacogenetics of cholesterol-lowering therapy.
Clin Chem Lab Med.2003;41:581-589.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=12747606&dopt=Abstract
Google Scholar 2.Evans WE, McLeod HL. Pharmacogenomics—drug disposition, drug targets, and side effects.
N Engl J Med.2003;348:538-549.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=12571262&dopt=Abstract
Google Scholar 3.Roses AD. Genome-based pharmacogenetics and the pharmaceutical industry.
Nat Rev Drug Discov.2002;1:541-549.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=12120260&dopt=Abstract
Google Scholar 4.Ordovas JM, Mooser V. The
APOE locus and the pharmacogenetics of
lipid response.
Curr Opin Lipidol.2002;13:113-117.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11891412&dopt=Abstract
Google Scholar 5.Brorholt-Petersen JU, Jensen HK, Raungaard B, Gregersen N, Faergeman O. LDL-receptor gene mutations and the hypocholesterolemic response to
statin therapy.
Clin Genet.2001;59:397-405.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11453971&dopt=Abstract
Google Scholar 6.Kuivenhoven JA, Jukema JW, Zwinderman AH.
et al. Regression Growth Evaluation Statin Study Group. The role of a common variant of the cholesteryl ester transfer protein
gene in the progression of coronary atherosclerosis.
N Engl J Med.1998;338:86-93.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9420339&dopt=Abstract
Google Scholar 7.Ordovas JM, Lopez-Miranda J, Perez-Jimenez F.
et al. Effect of apolipoprotein E and A-IV phenotypes on the low density lipoprotein
response to HMG-CoA reductase inhibitor therapy.
Atherosclerosis.1995;113:157-166.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7605354&dopt=Abstract
Google Scholar 8.Hansen PS, Meinertz H, Gerdes LU, Klausen IC, Faergeman O. Treatment of patients with familial defective apolipoprotein B-100
with pravastatin and gemfibrozil: a two-period cross-over study.
Clin Investig.1994;72:1065-1070.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7711417&dopt=Abstract
Google Scholar 9.Watanabe J, Kobayashi K, Umeda F.
et al. Apolipoprotein E polymorphism affects the response to pravastatin on
plasma apolipoproteins in diabetic patients.
Diabetes Res Clin Pract.1993;20:21-27.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8344125&dopt=Abstract
Google Scholar 10.Tsimikas S, Mooser V. Molecular Biology of Lipoproteins and Dislipidemias
in Molecular Basis of Cardiovascular Disease. 2nd ed. Antman EM, ed. Philadelphia, Pa: Saunders; 2002.
11.Albert MA, Danielson E, Rifai N, Ridker PM. Effect of statin therapy on C-reactive protein levels: the Pravastatin
Inflammation/CRP Evaluation (PRINCE): a randomized trial and cohort study.
JAMA.2001;286:64-70.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11434828&dopt=Abstract
Google Scholar 12.Gotto AM, Farmer JA. Dyslipoproteinemias/Atherosclerosis: Pharmacologic
Therapy in Cardiovascular Therapeutics. 2nd ed. Antman EM, ed. Philadelphia, Pa: Saunders; 2002.
14.Kokoris M, Dix K, Moynihan K.
et al. High-throughput SNP genotyping with the Masscode system.
Mol Diagn.2000;5:329-340.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11172497&dopt=Abstract
Google Scholar 15.Westfall PH, Young SS. Resampling-based Multiple Testing. New York, NY: John Wiley & Sons; 1993.
16.Templeton AR, Boerwinkle E, Sing CF. A cladistic analysis of phenotypic associations with haplotypes inferred
from restriction endonuclease mapping, I: basic theory and an analysis of
alcohol dehydrogenase activity in Drosophila.
Genetics.1987;117:343-351.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=2822535&dopt=Abstract
Google Scholar 17.Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population
data.
Am J Hum Genet.2001;68:978-989.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11254454&dopt=Abstract
Google Scholar 18.Clement M, Posada D, Crandall KA. TCS: a computer program to estimate gene genealogies.
Mol Ecol.2000;9:1657-1659.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11050560&dopt=Abstract
Google Scholar 19.Pruitt KD, Maglott DR. RefSeq and LocusLink: NCBI gene-centered resources.
Nucleic Acids Res.2001;29:137-140.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11125071&dopt=Abstract
Google Scholar 20. Executive summary of the third report of the National Cholesterol Education
Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High
Blood Cholesterol in Adults (Adult Treatment Panel III).
JAMA.2001;285:2486-2497.http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&Dopt=r&uid=entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11368702&dopt=Abstract
Google Scholar