| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrine Care |
Department of Epidemiology and Biostatistics, University of California, San Francisco, California 94105
Address all correspondence and requests for reprints to: Dr. Roland D. Chapurlat, INSERM U403, Service de Rhumatologie, Pavillon F, Hôpital E. Herriot, 5 place d’Arsonval, 69437 Lyon, Cedex 03, France. E-mail: roland.chapurlat{at}laposte.net
Abstract
The efficacy of estrogen therapy may be modified in women who smoke because of increased catabolism of estrogen and the interaction of tobacco products with the estradiol receptor. We examined whether the efficacy of raloxifene differed in smoking vs. nonsmoking women. We compared change in bone mineral density and biochemical markers of bone turnover, and incidence of new vertebral fracture in postmenopausal women of the Multiple Outcomes on Raloxifene Efficacy trial, who were randomized to either raloxifene (60 or 120 mg/d) or placebo.
In the 17% of women who were current smokers, we found, compared with nonsmokers, lowered baseline trochanter bone mineral density (0.540 vs. 0.557 g/cm2; P < 0.001) and serum osteocalcin (24.8 vs. 26.6 ng/liter; P < 0.001). Baseline urinary type I collagen breakdown products was increased among smokers (291.8 vs. 276.9 µmol/liter; P = 0.04). Body mass index was also lower in smokers (24.3 vs. 25.4; P < 0.001). After 6 months of treatment, there was no significant difference in reduction of bone turnover between smokers and nonsmokers. After 4 yr of treatment, the smoking-treatment interaction was not significant between smokers and nonsmokers for the percent increase in femoral neck bone mineral density (P = 0.25), trochanter bone mineral density (P = 0.24), and spine bone mineral density (P = 0.37). The smoking-treatment interaction for reduction in vertebral fracture risk was not significant either [odds ratio for fracture, 0.67 (0.45–0.98) for smokers and 0.56 (0.47–0.68) for nonsmokers; P = 0.44]. These results were not modified after stratification by tertiles of body mass index or when comparing heavy smokers vs. light smokers.
We conclude that smoking does not influence the antiosteoporotic effect of raloxifene. This may represent an advantage over estrogen replacement therapy.
IN SEVERAL STUDIES, the incidence of forearm, vertebral, and hip osteoporotic fracture was increased among postmenopausal smokers (1, 2, 3, 4). This increased bone fragility in smokers has been attributed to reduced bone mineral density (5, 6, 7, 8), greater bone loss at various skeletal sites (9, 10), decreased bone formation (11, 12), and increased bone resorption (7, 13). In contrast, a few investigators could not find a relationship between smoking and osteoporosis (14, 15, 16). Nevertheless, a recent meta-analysis concluded that cigarette smoking has a major deleterious effect on bone, including increased bone loss and hip fracture risk in postmenopausal current smokers (17).
The negative effect of cigarette smoking is probably multifactorial. Smoking is associated with early menopause (16, 18), lower body weight (1, 15), and decreased intestinal calcium absorption (9, 10). In addition, serum E2 levels are decreased among smokers because of reduced aromatization (19) and enhanced inactivation through 2- hydroxylation of E2 (20, 21). Furthermore, constituents of tobacco bind to the ER and compete with E2 (21). Bone formation is depressed, possibly as a result of increased cortisol synthesis (22) and direct toxic effects on bone cells (23). Moreover, increased concentrations of oxygen-derived free radicals and reduced levels of vitamins in current smokers may be involved in stimulation of the bone resorption process (24, 25). Even treatment of osteoporosis with estrogen replacement therapy seems to be hampered by cigarette smoking, as serum levels of E2 are decreased among treated current smokers, probably because of hepatic hypercatabolism (26), resulting in a lower proportion of smoking women having an increase in bone mineral density (BMD) in response to treatment (27).
Raloxifene is a selective ER modulator with estrogen antagonist activity on endometrial and breast tissue and agonist effects on the skeleton and lipid metabolism (28). Raloxifene competes with endogenous estrogens for binding to the ER and may either activate or block estrogen action. In postmenopausal women with osteoporosis, raloxifene increases BMD and reduces bone turnover and risk of vertebral fracture (29).
As smoking induces a reduction in serum E2 levels and can directly activate the ER, binding of raloxifene to the ER may be influenced, and thus its therapeutic efficacy modified, by smoking. We investigated this hypothesis in postmenopausal osteoporotic women of the Multiple Outcomes on Raloxifene Efficacy (MORE) randomized trial, examining the efficacy of raloxifene on BMD, bone turnover, and vertebral fracture incidence among smokers and nonsmokers. Many biological variables, such as hormones and markers of bone turnover, have been measured in this trial, so we were able to also explore mechanisms of action of smoking.
Subjects and Methods
Subjects
The MORE trial included 7705 women who were at least 2 yr postmenopausal and had osteoporosis, defined as low BMD or radiographically apparent fractures. Women were divided into 2 subgroups and then were randomized to either placebo or 1 of 2 dosages of raloxifene (60 or 120 mg/d). The two doses of raloxifene are pooled in the current analysis to increase the sample size, as the outcomes of interest did not differ between the groups. Women with a femoral neck or lumbar spine BMD t score below -2.5 were assigned to subgroup 1, and those who had low BMD and at least 1 or more moderate or severe vertebral fractures, 2 or more mild vertebral fractures, or at least 2 moderate fractures regardless of their BMD were included in subgroup 2. In the present analysis, these 2 subgroups were pooled. Definitions of different types of vertebral fractures as well as exclusion criteria are given in a previous report on the MORE study (29). Upon entry into the study, all women received daily supplements of 400 to 600 IU cholecalciferol and 500 mg calcium.
Several clinical variables were recorded at baseline, such as smoking
status, height, and body weight. Smokers were also stratified into
light smokers (<20 cigarettes/d) and heavy smokers (
20
cigarettes/d). Ex-smokers were considered as nonsmokers.
Biochemical measurements
Bone markers, including serum osteocalcin (ELSAOSTEO, CIS-Bio International, Gif sur Yvette, France) (30), serum bone-specific alkaline phosphatase (Tandem-R Ostase, Hybritech, San Diego, CA) (31), serum procollagen peptide type I (Procollagen-PICP RIA kit, DiaSorin, Inc., Stillwater, MN), and urinary type I collagen breakdown products (CTX) (CrossLaps, Osteometer Biotech A/S, Herlev, Denmark) (32, 33) were measured by Covance Central Labs in 2642 women. The first measurement was made on baseline samples, and follow-up measurements were performed after 6 months and 1, 2, and 3 yr. In this article we will refer only to baseline and 6 month measurements.
Serum 25-hydroxyvitamin D (25OHD) was measured with a RIA (25OHD RIA, DiaSorin, Inc.), as was serum PTH (N-tact PTH SP, DiaSorin, Inc.), by Covance Central Laboratories, Indianapolis, IN. Serum E2 was measured using a sequential RIA (Diagnostics Products, Los Angeles, CA), by Covance Central Laboratories. These hormone measurements were made only at baseline.
BMD
BMD was assessed using three different machines: Lunar (Milwaukee, WI), Hologic, Inc. (Bedford, MA), and Norland (White Plains, NY). Results were then standardized across machine types.
Assessment of fracture
Details on assessment of vertebral fracture have been previously reported (29). Briefly, women had vertebral radiographs at baseline and 24, 36, and 48 months. When symptoms of vertebral fracture occurred, women underwent radiography at interim 6-month visits. Radiographs were evaluated by a radiologist blinded to the group assignment using a semiquantitative method.
Statistical analysis
We analyzed baseline characteristics in smokers and nonsmokers
using a t test for normally distributed continuous
variables, a Wilcoxon rank-sum test for nonnormally distributed
continuous variables, and a 
2 test for
categorical variables. In this post-hoc analysis, we
examined three outcomes: percent change from baseline in the four
markers of bone turnover after 6 months of treatment, percent change
from baseline in BMD after 4 yr of treatment, and occurrence of
incident vertebral fracture after 4 yr of treatment. The distributions
of the percent changes in the four markers of bone turnover were not
normally distributed and could not be log-transformed because markers
of percent changes are negative numbers. This precluded the use of
linear regression. Therefore, we dichotomized these variables at the
median change, and then used logistic regression to determine the odds
ratio for raloxifene vs. placebo for being above the median
percent change for a given marker of bone turnover in smokers and
nonsmokers. Then the smoking-treatment interaction was introduced in
each model to test whether the efficacy of raloxifene to reduce markers
of bone turnover was significantly different between smokers and
nonsmokers. We used t tests to examine the efficacy of
raloxifene to increase the mean BMD in smokers and nonsmokers, and then
we entered the smoking-treatment interaction into linear regression
models to examine whether the treatment efficacy was significantly
different between smokers and nonsmokers. Moreover, we examined the
odds that the reduction in vertebral fracture incidence due to
raloxifene treatment was different among smokers and nonsmokers, again
using the smoking-treatment interaction. For all of these analyses we
also performed stratification according to tertiles of body mass index
(BMI), as BMI might confound, at least in part, the association between
cigarette smoking and osteoporosis. Finally, these same analyses were
repeated among light and heavy smokers to explore a potential dose
effect. Statistical tests were two-sided. All analyses were conducted
using the SAS 6.12 software (SAS Institute, Inc., Cary,
NC).
Results
At baseline, 17% of the participants were current smokers. The
proportion of women with baseline undetectable E2 was similar in
smokers (49%) and nonsmokers (51%; P = 0.31).
Compared with nonsmokers, smokers had lower serum osteocalcin, higher
urinary CTX, and lower trochanteric and whole body BMD (Table 1
). Smokers were also thinner, younger,
and had lower values of serum 25OHD and PTH than nonsmokers. All other
baseline variables were similar between smokers and nonsmokers (Table 1).
|
). Similarly, the percent increases in
femoral neck, trochanter, lumbar spine, and whole body BMD due to
raloxifene treatment were not significantly different among smokers and
nonsmokers, and the smoking-treatment interaction was not significant
(Table 3). In the placebo group, smokers
tended to lose bone more rapidly than nonsmokers, but this was not
statistically significant (Table 3). Raloxifene reduced the incidence
of vertebral fracture similarly in smokers [odds ratio, 0.67
(0.45–0.98)] and nonsmokers [odds ratio, 0.56 (0.47–0.68)], and
the smoking-treatment interaction term was not significant
(P = 0.44).
|
|
20 cigarettes/d) smokers regarding the efficacy of raloxifene on
markers of bone turnover, increase in BMD, and reductions in the
incidence of vertebral fracture (data not shown). Discussion
We found that postmenopausal osteoporotic women who smoked had somewhat decreased bone formation, increased bone resorption, and lowered BMD at baseline compared with women who did not smoke. There was no difference in the efficacy of raloxifene, however, to reduce bone turnover, increase BMD, or decrease vertebral fracture incidence between smokers and nonsmokers. These results were not modified after stratification by tertiles of BMI and after taking into account the number of cigarettes smoked.
Our findings confirm previously published data indicating that postmenopausal women who smoke have lower bone mass than nonsmokers, and thus have a greater risk of fracture (17). The uncoupling of bone turnover, as shown by decreased serum osteocalcin and increased urinary CTX, could explain this bone loss. Women of the MORE trial who were current smokers also had reduced levels of serum 25OHD, which are usually explained by lower dietary intakes of vitamin D (34). Low intakes of vitamin D are generally associated with low dietary intakes of calcium (35). This is likely to be another contributor to the lower bone mass of smokers. Nonetheless, one could have expected increased serum PTH concentrations in response to lowered 25OHD, whereas in our study serum PTH was greater in nonsmokers. The decreased level of serum PTH among smokers, which has already been described in several studies (35, 36, 37), might be due to direct toxicity for the parathyroid gland of certain compounds present in tobacco smoke (35). Furthermore, in our analysis, smokers had an increased bone resorption rate, so that the increased calcium flux from their skeleton is likely to have participated in these lowered PTH concentrations.
The efficacy of raloxifene to prevent the risk of incident vertebral fracture was not significantly different between smokers and nonsmokers. Moreover, there were no significant smoking-treatment interactions for the reduction of bone turnover or for the increase in BMD. Hence, we can assume that tobacco products do not notably influence the effect of raloxifene on bone. This might be explained by equivalent concentrations of serum E2 between smokers and nonsmokers at baseline, so that E2 and raloxifene competed similarly to bind to the ER. Indeed, in contrast to other studies, we did not find reduced concentrations of serum E2 among smokers. In addition, we found no evidence that the direct binding to the ER by some tobacco products (21) has the capability to substantially impede binding of raloxifene to the ER. In contrast, the efficacy of estrogen replacement therapy is probably diminished among smokers, as their serum concentrations of E2 are reduced when treated (26, 38), and the proportion of postmenopausal women taking estrogen who fail to increase their BMD is significantly greater in smokers than in nonsmokers (27). Accordingly, the level of bone turnover of women receiving estrogen replacement therapy is significantly more decreased in nonsmokers than in smokers (39). Reduced concentrations of serum E2 were observed, however, only in smoking women taking oral estrogen, not in those receiving transdermal estrogen (40), possibly because of a first pass effect. Data concerning the BMD increase and the bone turnover decrease among smokers in response to transdermal estrogen are lacking. Therefore, raloxifene might have an advantage over estrogen replacement therapy, at least by the oral route, in postmenopausal women who smoke.
As BMI is a significant predictor of bone mass and fracture and is associated with smoking, our analyses were repeated within each tertile of BMI to detect a potential influence of smoking on treatment efficacy in women who had the lowest BMI. Thus, we found no interaction between smoking and treatment efficacy for any of the outcomes we studied. Thus, even women who smoke and are among the thinnest benefit from treatment with raloxifene. On the other hand, to examine the potential dose-effect of smoking, we also determined whether there was any significant smoking-treatment interaction among light and heavy smokers. We found that the bone effect of raloxifene was similar among heavy smokers and light smokers, so we can rule out a decreased efficacy in the subgroup of postmenopausal women who smoke at least 20 cigarettes/d.
Our study has some limitations, as it is a post-hoc analysis of a randomized trial. The power to detect a difference in antifracture efficacy of raloxifene between smokers and nonsmokers may have been limited, as only 17% of participants were current smokers. Power, however, was greater with continuous variables such as markers of bone turnover and BMD, and as for fracture incidence, changes in these variables were not influenced by smoking. The generalizability of our results is limited to Caucasian postmenopausal women with osteoporosis.
We conclude that postmenopausal osteoporotic women who smoke have lower BMD than nonsmokers, are thinner, and have reduced bone formation and increased bone resorption. Nevertheless, raloxifene had the same antiosteoporotic efficacy in smokers as in nonsmokers, and this may represent an advantage over estrogen replacement therapy.
Acknowledgments
Footnotes
Abbreviations: BMD, Bone mineral density; BMI, body mass index; CTX, type I collagen breakdown products; MORE, Multiple Outcomes on Raloxifene Efficacy; 25OHD, 25-hydroxyvitamin D.
Received March 22, 2001.
Accepted May 17, 2001.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
