Koo et al.1
recently reported that the frequency of the 27-base pair repeat polymorphism
in intron 4, a noncoding part of the endothelial nitric oxide synthase (eNOS)
gene, was higher in Korean patients with idiopathic osteonecrosis of the hip
than in controls (9% compared with 2.4%; p = 0.03). Since this 4a polymorphism
is associated with reduced synthesis of eNOS, those authors concluded that a
carrier state of the 4a allele in intron 4 might be a genetic risk factor for
osteonecrosis of the femoral head and could provide insight into the
protective role of nitric oxide in the pathogenesis of this condition. In
their study, Koo et al. determined that the frequency of the intron 4
polymorphism did not differ when patients with osteonecrosis of the hip
secondary to steroids or alcohol were compared with controls.
The pathogenesis of osteonecrosis probably reflects multiple
etiologies2. A
previously postulated sequence of venous thrombosis with outflow obstruction
mediated by thrombophilia and/or hypofibrinolysis, leading to increased
intraosseous venous pressure, reduced arterial flow, and hypoxia appears to be
important in the development of
osteonecrosis2-5.
We speculate that an eNOS-polymorphism-mediated reduction in nitric oxide
production plays a role in osteonecrosis by diminishing femoral head
perfusion, either independently or in addition to the effect of thrombophilia
and
hypofibrinolysis6.
Impairment of nitric oxide production by the T-786C
eNOS7,8
polymorphism promotes vasoconstriction and platelet activation, recruitment,
and aggregation in human
subjects9 and
diminishes
angiogenesis10,11
and bone formation in stem cell and animal
models12. In the
present study, we chose to examine the T-786C eNOS polymorphism because it is
associated with arterial and venous disease in Buerger
disease13. We
speculate that
eNOS-mediated1
impairment of nitric oxide production is associated with idiopathic
osteonecrosis of the head of the femur. A thymine replacement by cytosine at
nucleotide —786 of the eNOS gene is responsible for a reduction in the
eNOS gene-promoter activity and for reduced nitric oxide
production14,15.
We assessed relationships between the T-786C eNOS polymorphism and
osteonecrosis of the head of the femur in order to understand better its
pathophysiology, with the ultimate goal of developing new medical therapies
for osteonecrosis.
The present prospective research study was approved by the institutional
review board at the Jewish Hospital in Cincinnati, Ohio; signed informed
consent for participation in the study was obtained from each patient.
Study Design
A consecutive series of ninety-five patients (148 hips) with osteonecrosis
was studied prospectively. Patients were referred by orthopaedic surgeons in
six Midwestern states for coagulation tests when the etiology of the
osteonecrosis was typically not known. This may have accounted for a larger
proportion of patients with idiopathic osteonecrosis than is usually seen by
orthopaedists and rheumatologists. To characterize the disease as idiopathic,
we excluded patients with recognized
causes6 of secondary
osteonecrosis:
alcoholism16,
corticosteroid use (approximately 6000 mg of prednisolone or its equivalent
over one to sixteen
months)17,
connective-tissue
disease4 (including
antiphospholipid antibody syndrome), previous use of
chemotherapy18, HIV
or AIDS (human immunodeficiency virus or acquired immune deficiency
syndrome)19, and
hip joint trauma
(dislocation20 or
fracture21).
Thirty-six nonsmoking patients (fifty-five hips) were identified as having
idiopathic osteonecrosis (Fig.
1). An additional eighteen patients (thirty hips) who smoked at
least one pack of cigarettes per day also were identified on the basis of
conventional criteria (i.e., the absence of recognized causes of secondary
osteonecrosis)6,22
as having idiopathic osteonecrosis (Fig.
2), but, for the purposes of analysis, these eighteen smokers
(thirty hips) were added to the group of forty-one patients (sixty-three hips)
with
conventional6,22
causes of secondary osteonecrosis (Fig.
1). Analyses were also done with all ninety-five patients (148
hips) who had osteonecrosis being pooled together in a single cohort
(Fig. 3).
Each patient had to qualify as having osteonecrosis of the head of the
femur on the basis of a thorough history and physical examination,
anteroposterior and frog-leg lateral radiographs of both hips, and magnetic
resonance
imaging23. Magnetic
resonance imaging was used to help to verify the clinical diagnosis of
osteonecrosis; no attempt was made to quantify the extent of femoral head
involvement with use of magnetic resonance imaging.
None of the patients with osteonecrosis had historical or clinical evidence
of coronary
vasospasm24 or
Buerger disease13,
coexisting disorders that may be associated with the T-786C eNOS polymorphism.
None of the patients with osteonecrosis or controls were organ-transplant
recipients.
Seventy-two healthy normal adult controls were matched to patients on the
basis of race, gender, and age; this control group included forty previously
described healthy adult hospital
personnel25,26
and thirty-two healthy subjects who were evaluated during family studies of
hyperlipidemic patients. A detailed medical history was obtained for all
seventy-two controls, none of whom had osteonecrosis or clinical symptoms
suggesting osteonecrosis of the femoral head. We cannot exclude the remote
possibility (approximately 0.0026 patient per 100
person-years)27
that any of the seventy-two healthy adult normal controls had asymptomatic,
subclinical osteonecrosis or that it would develop later. We were able to
match controls according to race, gender, and age to all forty-one patients
with secondary osteonecrosis and to fifty-one of fifty-four patients with
idiopathic osteonecrosis.
Laboratory Methods
As previously
described26, after
an overnight fast, blood for polymerase chain reaction analysis was drawn in
tubes containing ethylene diaminetetraacetic acid (EDTA) and the
deoxyribonucleic acid (DNA) was extracted for subsequent analysis of the
T-786C eNOS
polymorphism14,24,28,29.
DNA was isolated with the Capture Column (Gentra Systems, Minneapolis,
Minnesota). Polymerase chain reaction measures were made of the T-786C eNOS
polymorphism14,24,28,29.
One hundred nanograms of patient DNA was denatured at 95°C for five
minutes and then at thirty-one cycles of 95°C for one minute, 60°C for
one minute, and then 72°C for one minute. The product was digested with
Tth111 I (New England Biolabs, Beverly, Massachusetts). The forward primer for
the eNOS polymorphism was 5-tgg aga gtg ctg gtg tacc cca-3. The reverse primer
was 5-gcc tcc acc ccc acc ctg
tg-329. One hundred
nanograms of patient DNA was denatured at 95°C for five minutes and then
at thirty-two cycles of 94°C for 0.5 minute, 63°C for 0.5 minute, and
then 72°C for 0.5 minute. The product was digested with Msp I (New England
Biolabs). Products of the polymerase chain reactions were then electrophoresed
on a 10% polyacrylamide gel, and the bands were visualized with ethidium
bromide.
Statistical Methods
Categorical comparisons between patients with idiopathic and secondary
osteonecrosis and between patients with osteonecrosis and race, gender, and
age-matched controls were assessed with use of chi-square tests for 2 ×
3 tables (Figs. 1,
2, and
3). Fisher exact tests were
used when one cell size was =5.
Odds ratios with 95% confidence intervals were calculated for the
comparison between patients and controls with regard to eNOS genotypes (TT
[homozygous eNOS mutant genotype], CT [heterozygous mutant genotype], and CC
[homozygous wild-type normal]).
Five separate stepwise logistic regression models were run with the
following groups (all with seventy-two controls) as the dependent variable:
(1) thirty-six nonsmoking patients with idiopathic osteonecrosis, (2)
fifty-four patients with idiopathic osteonecrosis (including thirty-six
nonsmokers and eighteen smokers), (3) forty-one patients with secondary
osteonecrosis, (4) fifty-nine patients with secondary osteonecrosis (including
forty-one patients with
conventional6,22
secondary osteonecrosis and eighteen smokers in whom the osteonecrosis was
otherwise idiopathic), and (5) all ninety-five patients with osteonecrosis.
Explanatory variables included eNOS genotype, age, gender, race, and (with the
exception of the thirty-six nonsmoking patients with idiopathic osteonecrosis)
smoking. Additionally, four separate stepwise logistic regression models were
run with the following groups (all with seventy-two controls) as the dependent
variable: (1) fifty-four patients with idiopathic osteonecrosis (including
thirty-six nonsmokers and eighteen smokers), (2) forty-one patients with
secondary osteonecrosis, (3) fifty-nine patients with secondary osteonecrosis
(including forty-one patients with
conventional6,22
secondary osteonecrosis and eighteen smokers in whom the osteonecrosis was
otherwise idiopathic), and (4) all ninety-five patients with osteonecrosis.
Explanatory variables in these four logistic regression models included eNOS
genotype, age, gender, race, and smoking as well as a smoking-eNOS genotype
interaction term, which represented the product of smoking (coded as 1 for no
smoking and 2 for smoking) and eNOS genotype (coded as 1 for CC, 2 for CT, and
3 for TT).
Patients and Controls
Demographic characteristics of the patients and controls are summarized in
Table I. Most patients were
white, with the mean age (and standard deviation) of the cohort being 48
± 10 years. A majority of patients with idiopathic osteonecrosis were
men.
As shown in Figure 1,
homozygosity for the mutant eNOS allele (TT) was present in eight (22%) of
thirty-six nonsmoking patients with idiopathic osteonecrosis as compared with
one (3%) of thirty-six controls, heterozygosity for the mutant allele (TC) was
present in nineteen patients (53%) as compared with ten controls (28%), and
the wild-type normal genotype (CC) was present in nine patients (25%) as
compared with twenty-five controls (69%) (chi square = 15.8, degrees of
freedom = 2, p = 0.0004). The eNOS mutant allele frequency in patients with
idiopathic osteonecrosis (49%; thirty-five of seventy-two) was greater than
that in race, gender, and age-matched controls (17%; twelve of seventy-two) (p
< 0.0001). When the patients and controls were compared with regard to the
eNOS TT genotype (homozygous eNOS mutant genotype) and the eNOS CC genotype
(homozygous wild-type normal), the odds ratio was 22.2 (95% confidence
interval, 2.4 to 203.4). When the patients and controls were compared with
regard to the eNOS TC genotype (heterozygous eNOS mutant genotype) and the CC
genotype, the odds ratio was 5.3 (95% confidence interval, 1.79 to 15.5).
Logistic regression revealed that, for the thirty-six nonsmoking patients with
idiopathic osteonecrosis, the T-786C eNOS mutant allele was independently
associated with idiopathic osteonecrosis (odds ratio, 6.0; 95% confidence
interval, 2.51 to 14.4).
Of the forty-one patients with
conventional6,16,17,22
causes of secondary osteonecrosis, sixteen had a history of long-term,
high-dose corticosteroid use; four had a history of alcoholism; two had a
history of both alcoholism and corticosteroid use; and nineteen had a history
of connective tissue-antiphospholipid antibody diseases. Of the forty-one
patients with secondary osteonecrosis, nine were smokers. When the eighteen
smokers with otherwise idiopathic osteonecrosis were added to the group of
forty-one patients with secondary osteonecrosis, the distribution of the
T-786C eNOS mutant allele did not differ between these fifty-nine patients
with secondary osteonecrosis and fifty-two race, gender, and age-matched
controls (p = 0.19) (Fig. 1).
The T-786C eNOS mutant allele frequency in patients with secondary
osteonecrosis (30%; thirty-five of 118) was not different from that in
controls (21%; 22 of 104) (p = 0.15) (Fig.
1). Logistic regression revealed that smoking was more common in
patients with secondary osteonecrosis than in controls (odds ratio, 9.16; 95%
confidence interval, 3.30 to 25.4). In the separate logistic regression model
for secondary osteonecrosis, with a smoking-eNOS genotype interaction term as
an additional explanatory variable, the smoking-eNOS interaction term was not
significantly associated with secondary osteonecrosis (p = 0.09).
The thirty-six nonsmoking patients with idiopathic osteonecrosis differed
from the fifty-nine patients with secondary osteonecrosis in that they were
more likely to have mutant eNOS genotypes (chi square = 6.8, degrees of
freedom = 2, p = 0.033) and to have a higher mutant T allele frequency (49%
compared with 30%; p = 0.009) (Fig.
1).
As shown in Figure 2, TT
eNOS homozygosity was present in eleven (20%) of fifty-four patients with
idiopathic osteonecrosis (including thirty-six nonsmokers and eighteen
smokers) as compared with one (2%) of fifty-one controls, TC heterozygosity
was present in twenty-eight patients (52%) as compared with seventeen controls
(33%), and the wild-type normal genotype (CC) was present in fifteen patients
(28%) as compared with thirty-three controls (65%) (chi square = 17.7, degrees
of freedom = 2, p = 0.0001). When the patients and controls were compared with
regard to the eNOS TT genotype (homozygous eNOS mutant genotype) and the eNOS
CC genotype (homozygous wild-type normal), the odds ratio was 24.2 (95%
confidence interval, 2.86 to 204.9). When the patients and controls were
compared with regard to the eNOS TC genotype (heterozygous eNOS mutant
genotype) and the CC genotype, the odds ratio was 3.62 (95% confidence
interval, 1.54 to 8.54. When the patients and controls were compared with
regard to the eNOS TT genotype and combined TC and CC genotypes, the odds
ratio was 12.8 (95% confidence interval, 1.59 to 103.1). The T-786C eNOS
mutant allele frequency in patients with idiopathic osteonecrosis (46%; fifty
of 108) was higher than that in race, gender, and age-matched controls (19%;
nineteen of 102) (p < 0.0001) (Fig.
2). Logistic regression revealed that, in the fifty-four patients
with idiopathic osteonecrosis, the T-786 eNOS mutant allele was independently
associated with idiopathic osteonecrosis (odds ratio, 5.86; 95% confidence
interval, 2.63 to 13.1). Moreover, the smoking-eNOS interaction term was
positively associated with idiopathic osteonecrosis (p < 0.0001, odds
ratio, 3.89; 95% confidence interval, 2.10 to 7.21). The significant
interaction term denotes a combined, synergistic, and not simply additive
effect of the two explanatory variables (the T-786C eNOS mutant allele and
smoking) for the development of idiopathic osteonecrosis.
The distribution of the T-786C eNOS mutant allele did not differ between
forty-one patients with secondary osteonecrosis and race, gender, and
age-matched controls (p = 0.52) (Fig.
2). The T-786C eNOS mutant allele frequency in patients with
secondary osteonecrosis (24%; twenty of eighty-two) was not different from
that in controls (23%; nineteen of eighty-two) (p = 0.85)
(Fig. 2). Logistic regression
revealed that the only significant explanatory variable for secondary
osteonecrosis was cigarette smoking, which was more common in patients (odds
ratio, 3.59; 95% confidence interval, 1.14 to 11.26). In the separate logistic
regression model for secondary osteonecrosis, with a smoking-eNOS genotype
interaction term as an additional explanatory variable, the smoking-eNOS
interaction term was not significantly associated with secondary osteonecrosis
(p > 0.15).
The fifty-four patients with idiopathic osteonecrosis differed from
forty-one patients with secondary osteonecrosis in that they were more likely
to have mutant eNOS genotypes (chi square = 9.72, degrees of freedom = 2, p =
0.008) and to have a higher mutant T allele frequency (p = 0.002)
(Fig. 2).
As shown in Figure 3, when
the fifty-four patients with idiopathic osteonecrosis were grouped together
with the forty-one patients with secondary osteonecrosis, TT eNOS homozygosity
was present in fourteen (15%) of ninety-five patients with osteonecrosis as
compared with one (1%) of seventy-two controls, TC heterozygosity was present
in forty-two patients (44%) as compared with twenty-seven controls (38%), and
the wild-type normal genotype (CC) was present in thirty-nine patients (41%)
as compared with forty-four controls (61%) (chi square = 11.9, degrees of
freedom = 2, p = 0.0026). When the full cohort of ninety-five patients was
compared with the seventy-two controls with regard to the eNOS TT genotype
(homozygous eNOS mutant genotype) and the eNOS CC genotype (homozygous
wild-type normal), the odds ratio was 15.8 (95% confidence interval, 1.98 to
125.7). When the patients and controls were compared with regard to the eNOS
TC genotype (heterozygous eNOS mutant genotype) and the CC genotype, the odds
ratio was 1.76 (95% confidence interval, 0.92 to 3.35). When the patients and
controls were compared with regard to the eNOS TT genotype and combined TC and
CC genotypes, the odds ratio was 12.27 (95% confidence interval, 1.57 to
95.7). The frequency of the T-786C eNOS mutant allele in the pooled group of
patients with osteonecrosis (37%; seventy of 190) was significantly higher
than that in controls (20%; twenty-nine of 144) (p = 0.0009)
(Fig. 3). In the stepwise
logistic regression model for all ninety-five patients with osteonecrosis,
smoking (odds ratio, 5.26; 95% confidence interval, 1.89 to 14.69) and eNOS
genotype (odds ratio, 2.75; 95% confidence interval, 1.54 to 4.90) were
associated with osteonecrosis. In the separate logistic regression model, with
a smoking-eNOS genotype interaction term as an additional explanatory
variable, the smoking-eNOS interaction term was positively associated with
osteonecrosis (odds ratio, 2.61; 95% confidence interval, 1.62 to 4.19). The
significant interaction term denotes a combined, synergistic, and not simply
additive effect of the two explanatory variables (the T-786C eNOS mutant
allele and smoking) for the development of osteonecrosis.
When all twenty-seven cigarette smokers were pooled together as a different
secondary osteonecrosis group, there were no differences between those
individuals and twenty-five race, gender, and age-matched controls with regard
to TT homozygosity (p = 0.61, Fisher exact test) or the frequency of the
mutant T allele (31% [seventeen of fifty-four] compared with 24% [twelve of
fifty]; p = 0.4).
When the sixteen patients who had a history of corticosteroid use were
pooled together as a different secondary osteonecrosis group, there were no
differences between those individuals and sixteen race, gender, and
age-matched controls with regard to TT homozygosity (p = 0.48, Fisher exact
test) or the frequency of the mutant T allele (31% [ten of thirty-two]
compared with 25% [eight of thirty-two]; p = 0.58).
When the nineteen patients with connective-tissue-antiphospholipid antibody
disease were pooled together as a different secondary osteonecrosis group,
there were no differences between those individuals and eighteen race, gender,
and age-matched controls with regard to TT homozygosity (p = 1.0, Fisher exact
test) or the frequency of the mutant T allele (24% [nine of thirty-eight]
compared with 14% [five of thirty-six]; p = 0.28).