Early smoking is associated with peak bone mass and prevalent fractures in young, healthy men

Abstract

Smoking is associated with lower areal bone mineral density (aBMD) and higher fracture risk, although most evidence has been derived from studies in elderly subjects. This study investigates smoking habits in relation to areal and volumetric bone parameters and fracture prevalence in young, healthy males at peak bone mass. Healthy male siblings (n = 677) at the age of peak bone mass (25 to 45 years) were recruited in a cross-sectional population-based study. Trabecular and cortical bone parameters of the radius and cortical bone parameters of the tibia were assessed using peripheral quantitative computed tomography (pQCT). Areal bone mass was determined using dual energy X-ray absorptiometry (DXA). Sex steroids and bone markers were determined using immunoassays. Prevalent fractures and smoking habits were assessed using questionnaires. Self-reported fractures were more prevalent in the current and early smokers than in the never smokers (p < .05), with a fracture prevalence odds ratio for early smokers of 1.96 (95% confidence interval 1.18–3.24) after adjustment for age, weight, educational level, and alcohol use and exclusion of childhood fractures. Current smoking was associated with a larger endosteal circumference (β = 0.027 ± 0.009, p = .016) and a decreased cortical thickness (β = −0.034 ± 0.01, p = .020) at the tibia. In particular, early smokers (≤16 years) had a high fracture risk and lower areal BMD, together with a lower cortical bone area at the tibia and lower trabecular and cortical bone density at the radius. An interaction between free estradiol and current smoking was observed in statistical models predicting cortical area and thickness (β = 0.29 ± 0.11, p = .01). In conclusion, smoking at a young age is associated with unfavorable bone geometry and density and is associated with increased fracture prevalence, providing arguments for a disturbed acquisition of peak bone mass during puberty by smoking, possibly owing to an interaction with sex steroid action. © 2010 American Society for Bone and Mineral Research

Introduction

Smoking is a common lifestyle factor with a major impact on mortality rates owing to increased prevalence of cardiovascular disease, chronic obstructive lung disease, and cancer in smokers. In addition, substantial evidence has accumulated relating tobacco use to low bone mass and increased fracture risk.1, 2 In a large meta-analysis, smoking in older men and women was associated with a significantly increased risk of fracture, including osteoporotic fractures. The fracture risk associated with smoking was significantly higher in men than in women.1 In older men, increased bone resorption, unmatched by bone formation, was found in smokers, leading to lower bone mass and vertebral deformities.3

In young men, data on smoking are scarce, and the effects of smoking on peak bone mass are undefined. A recent report demonstrated lower bone mineral density (BMD) and smaller cortical thickness in young male smokers (18 to 19 years of age), although the mechanism by which smoking influences bone metabolism remains elusive, especially in young men.4

Sex steroid hormones play an important role in bone development and maintenance of bone mass during aging. In men, consistent associations between serum sex steroid levels and BMD, bone size, and fracture prevalence were observed.5–12 Previously, our group established an association between low bone mass and a perturbed sex steroid status, with lower estradiol concentrations in men with idiopathic osteoporosis, as well as their first-degree relatives.13, 14 Antiestrogenic effects of smoking have been recognized in women,15 although the data in men are limited. We hypothesize that part of the unfavorable effects of smoking on bone could be due to interaction of smoking with sex steroid action. However, smoking is also part of a set of lifestyle habits and social background that all could influence bone mass and body composition, apart from direct metabolic and endocrine effects associated with smoking, possibly influencing bone metabolism.

In this study we aim to investigate volumetric bone density and bone geometry using peripheral quantitative computed tomography (pQCT) in relation to smoking habits, sex steroids, and prevalent fractures in a sibling-pair setting of young, healthy male subjects, taking into account possible confounding lifestyle factors such as physical activity, alcohol and calcium intake, and level of education.

Materials and Methods

Study design and population

Participants were recruited from the population registries of three semirural to suburban communities around Ghent, Belgium. Men (n = 12,446) aged 25 to 45 years were contacted by direct mailing, briefly describing the study purpose and asking if they had a brother within the same age range also willing to participate (maximal age difference between brothers was set at 12 years). The overall response rate was 30.2%. Since we had no information on family structure from the population registry, no inclusion rate could be calculated. Most contacted men did not have an eligible brother, and finally, a sample of 768 young, healthy men agreed to participate who fulfilled the primary inclusion criterion of having a brother within the same age range. After exclusions, in total, 677 men were included in the study. Two-hundred ninety-six pairs of brothers (for a total of 592 men), in addition to 64 men as single participants, were included; when their brother could not participate in the study, 19 men were included as third brother in a family and 2 as fourth brother. All participants were in good health and completed questionnaires about previous illness and medication use. Exclusion criteria were defined as illnesses or medication use affecting body composition or hormone or bone metabolism—such as current or prolonged use of glucocorticosteroids, (anti)androgens, vitamin D supplements, insulin, or thyroxin; previous or current use of antiepileptic drugs; presence of hypogonadism, hyperthyroidism, cystic fibrosis, malabsorption, or eating disorders or disorders of collagen metabolism or bone development, and preence of chronic renal failure, alcohol abuse, or autoimmune rheumatoid disease. Calcium intake and alcohol use were noted from food recall data, and physical activity was scored using the questionnaire as proposed by Baecke and colleagues.16 Smoking behavior, number of cigarettes, starting age of smoking, education level, and prevalent bone fractures were recorded. In 11 subjects, starting age of smoking was incomplete. Early smoking was defined as starting to smoke at any age of 16 years or younger, based on the distribution of the starting age of smoking (Fig. 1) and bone growth during puberty.

Figure 1. Histograms illustrating the distribution of the starting age of smoking and the occurrence of fracture distribution according to age.

The study protocol was approved by the ethical committee of the Ghent University Hospital, and informed consent was obtained from all participants.

Areal BMD

Body weight and anthropometrics were measured in light indoor clothing without shoes. Standing height was measured using a wall-mounted Harpenden stadiometer (Holtain, Ltd., Crymuch, UK).

Bone mineral content (BMC) and areal bone mineral density (aBMD) at the lumbar spine and proximal femur (total hip region) were measured using dual X-ray absorptiometry (DXA) with a Hologic QDR-4500A device (Software Version 11.2.1, Hologic, Inc., Bedford, MA, USA). The coefficient of variation (CV, %) for both spine and whole-body calibration phantoms was less than 1%, as calculated from daily and weekly measurements, respectively.

Volumetric and geometric bone parameters (pQCT)

A pQCT device (XCT-2000, Stratec Medizintechnik, Pforzheim, Germany) was used to scan the dominant leg (tibia) and forearm (radius). The dominant side was selected to allow assessment of the relationship between muscle area and bone parameters. The cortical volumetric bone mineral density (vBMD, mg/cm³), cortical cross-sectional area (mm²), endosteal and periosteal circumferences, and cortical thickness (mm) were measured at the midradius (66% of bone length from the distal end) and tibia (66%). Trabecular vBMD was measured using a scan through the metaphysis (at 4% of the radius length) of the nondominant arm.

Biochemical determinations

Venous blood samples were obtained between 8 and 10 hours after overnight fasting. All serum samples were stored at −80°C until batch analysis. Commercial radioimmunoassays were used to determine serum levels of total testosterone, sex hormone–binding globulin (SHBG) (Orion Diagnostica, Espoo, Finland), and estradiol (Clinical Assay, DiaSorin s.r.l., Saluggia, Italy; modified protocol using a double amount of serum).17 Intact parathyroid hormone (PTH), C-terminal telopeptides of type I collagen (CTX), and procollagen type 1 amino-terminal propeptide (P1NP) were measured using an immunoelectrochemiluminescence technique (Modular, Roche Diagnostics, Mannheim, Germany). Free testosterone and free estradiol concentrations were calculated from serum total testosterone, estradiol, SHBG, and albumin concentrations using a previously validated equation derived from the mass-action law.18 Estradiol was determined using a modified protocol with a double amount of serum19 with a detection limit of 2.5 pg/mL (9.2 pmol/L). All samples of our participants had estradiol concentrations above this detection limit. Intra- and interassay CVs were below 10% and 15% for all measurements.

Statistics

Descriptives are expressed as mean ± standard deviation or median (first through third quartiles) when criteria for normality were not fulfilled (Kolmogorov-Smirnov), and variables (bone parameters and steroid concentrations) were log-transformed in subsequent linear models. Linear mixed-effects modeling with random intercepts and a simple residual correlation structure for both random and fixed effects was used to evaluate cross-sectional relationships in our study population, taking the interdependence of measurements within families into account. Logistic regression, taking into account the family structure, was modeled in S-Plus using the glme function.

Parameters of fixed effects were estimated via restricted maximum-likelihood estimation and reported as estimates of effect size (β) with their respective standard error.

Associations were considered significant at p values of less than .05. Statistical analyses were performed using S-Plus 7.0 (Insightful, Seattle, WA, USA) and SAS 9.1.3 software (SAS Institute, Inc., Cary, NC, USA).

Results

Baseline characteristics

Table 1 illustrates the anthropometric indices and hormonal serum concentrations in current, former, and never smokers. All 677 subjects were in good health, and hormonal concentrations were within the expected range. Twenty-three percent of our healthy males were current smokers, whereas 397 (59%) never had used tobacco. Current smokers were slightly younger than never or former smokers, and (free) testosterone concentrations were higher (adjusted for age, height, and weight) in the current smokers, whereas no difference in SHBG or (free) estradiol concentrations were observed. Calcium intake (−11%, p < .05), years of education (median difference −2 year, p < .001), and physical activity (−3%, p = .05) were lower in smokers. Figure 1 illustrates the starting age of smoking and reported fractures according to age. Mean age at which smoking started was 16.3 years, and the highest frequency of fractures occurred between the ages 14 and 16 years (see Fig. 1).

Table 1. Descriptive Anthropometric and Hormonal Parameters According to Smoking Behavior

	Never smoker (n = 397)	Current smoker (n = 155)	Former smoker (n = 125)
Data presented as mean ± SD or median (25th through 75th percentile). * p < .001; ** p < .01; *** p < .05 (smokers versus never smokers; hormonal comparisons are corrected for age, weight, and height).
Age (years)	34.5 ± 5.5	33.4 ± 5.5**	35.7 ± 5.5***
Height (cm)	180 ± 7	179 ± 6	179 ± 6
Weight (kg)	81 ± 12	82 ± 13	81 ± 10
Pack-years	0	11 (5–18)*	9 (5–14)*
BMI (kg/m^2)	24.6 ± 3.5	25.8 ± 4.1	25.4 ± 2.8
Years of education (years)	14.7 ± 2.9	13.8 ± 2.8*	13.6 ± 3.2**
Calcium intake (mg/24 h)	646 ± 296	577 ± 290**	598 ± 241
Alcohol use (drinks/week)	7 (3–13)	10 (6–17)*	8 (5–16)
Testosterone (ng/dL)	555 ± 139	603 ± 152*	558 ± 151
Free testosterone (ng/dL)	13.5 ± 3.1	14.6 ± 3.4*	13.7 ± 3.3
Estradiol (E2) (pg/mL)	20.2 ± 5.0	20.7 ± 4.3	20.0 ± 4.3
Free estradiol (E2) (pg/mL)	0.40 ± 0.10	0.41 ± 0.10	0.40 ± 0.08

Thirty-three percent (n = 219) of the subjects reported one or more clinically significant bone fractures, of which 16% (n = 108) sustained a wrist fracture. Other fractures included shoulder (n = 35; 5%), humerus (n = 23; 4%), and lower limb (n = 32; 7%) fractures. Four participants reported a hip fracture (0.6%) after major trauma. Finger, toe, and cranial fractures were excluded from the analysis. Most fractures were associated with sports activities, although the intensity of trauma could not be quantified.

In never smokers, 104 (26%) of the participants reported a significant fracture, 64 (42%) in current smokers, and 51 (41%) in former smokers (χ², p = .0004). Categorizing the smokers (current and former) in early (≤16 years) and late smokers (>16 years) demonstrated a significantly higher proportion of fractures in the early smokers (n = 72, 44%) and late smokers (n = 39, 38%) (χ² comparing never, early, and late smokers, p = .0002). In 11 subjects (4 with a fracture), no starting age of smoking was recorded. In early smokers, median pack-years (11, range 7–21) was higher than in former smokers (9, range 3–13).

Smoking in relation to the areal bone parameters (DXA)

Figure 2 illustrates the areal bone parameters according to smoking status and starting age of smoking. Early smokers had significant lower aBMD at the spine, hip, and total-body DXA measurements, whereas no significant differences in aBMD or aBMC were observed between current smokers or never smokers (data not shown). Significant negative associations were observed between number of pack-years smoked and aBMD at the spine (β = −0.001 ± 0.0005, p = .04), total hip (β = −0.001 ± 0.0005, p = .03), and total body (β = −0.001 ± 0.0004, p = .02), as well as total-body BMC (β = −0.01 ± 0.0003, p = .002). Projectional bone area was not related to the amount of tobacco previously smoked.

Figure 2. Areal BMD (g/cm²) and Bmc (g) at the lumbar spine, hip, and total body according to the smoking habits: never smokers, smoking started during puberty (≤16 years), and smoking started later than 16 years. p values adjusted for age, height, and weight.

Bone size by pQCT in relation to smoking

Bone size and geometry at the radius and tibia in relation to smoking habits are given in Table 2. At both the radius and the tibia, early smoking was associated with a larger endosteal circumference and, at the tibia, with a smaller cortical thickness and smaller cortical area. Except for cortical area at the tibia, these associations remained significant after adjustment for education level, alcohol use, physical activity, and calcium intake. Slightly larger periosteal circumferences at the radius and the tibia were observed in current smokers, with a marked smaller cortical thickness at the tibia in current and former smokers. Current smoking was associated with a larger endosteal circumference (β = 0.028 ± 0.008, p = .001) and a decreased cortical thickness (β = −0.043 ± 0.011, p = .0001) at the tibia. At the tibia, the number of pack-years was associated with a larger endosteal circumference (β = 0.001 ± 0.0004, p = .006) and a decreased cortical thickness (β = −0.002 ± 0.0005, p = .003), whereas no association was observed with periosteal circumference (p = .16). In order to discriminate the effects of early smoking versus total amount of pack-years smoked, we performed additional analyses in a subgroup of smokers with limited pack-years (<12 pack-years). In this subgroup, no significant difference in pack-years was noted between early and late smokers, whereas a decrease in cortical thickness (−4.8%, p = .012) and a wider endosteal cavity (+1.6%, p = .037) were still observed in early smokers.

Table 2. Volumetric Bone Mass Density and Bone Geometry at the Radius and Tibia, According to Current Smoking Habits

	Never smoker (n = 397)	Current smoker (n = 155)	Former smoker (n = 125)	Early smoker, ≤16 years (n = 165)	Late smoker, >16 years (n = 104)
Data presented as mean ± SD. * p < .001; ^**p < .01; *** p < .05 (smokers (current or former/ early or late) versus never smokers; adjusted for age, weight and height). Early and late smokers refer to the age started smoking and comprise current and former smokers.
Radius
Trabecular bone area (mm^2) at 4%	186 ± 25	188 ± 28	187 ± 25	188 ± 28	186 ± 25
Trabecular bone density (mg/cm^3) at 4%	231 ± 41	229 ± 39	221 ± 38^c	223 ± 38^b	228 ± 39
Cortical bone density (mg/cm^3)	1102 ± 34	1098 ± 37	1097 ± 35	1095 ± 36^c	1102 ± 35
Cortical bone area (mm^2)	101 ± 13	103 ± 15	101 ± 12	101 ± 14	103 ± 13
Periostal circumference (mm)	48.0 ± 3.7	48.8 ± 4.0^c	48.5 ± 3.5	48.7 ± 3.8	48.6 ± 3.7
Endostal circumference (mm)	32.1 ± 4.4	32.8 ± 4.6	32.9 ± 4.3	33.1 ± 4.5***	32.6 ± 4.3
Cortical thickness (mm)	2.52 ± 0.32	2.54 ± 0.32	2.49 ± 0.31	2.49 ± 0.33	2.55 ± 0.31
Tibia
Cortical bone density (mg/cm^3)	1114 ± 23	1112 ± 27	1109 ± 24	1108 ± 25	1112 ± 26
Cortical bone area (mm^2)	367 ± 50	362 ± 45	354 ± 42	355 ± 43*	363 ± 45
Periostal circumference (mm)	94.6 ± 6.1	95.4 ± 6.0***	94.4 ± 5.4	95.1 ± 5.9	95.0 ± 5.7
Endostal circumference (mm)	66.0 ± 6.6	67.4 ± 7.3*	66.8 ± 6.2	67.6 ± 6.7*	66.6 ± 6.8***
Cortical thickness (mm)	4.59 ± 0.57	4.46 ± 0.57^*	4.40 ± 0.50***	4.38 ± 0.52*	4.50 ± 0.56

Smoking in relation to the volumetric bone density and bone turnover markers

At the radius, early smokers had significantly lower trabecular and cortical volumetric bone densities (see Table 2). At the tibia, no effect of early, late, or current smoking on cortical density was observed (p > .05). The number of pack-years smoked was not related to cortical vBMD at the radius or the tibia, but a significant negative association was observed with trabecular density at the radius (β = −0.002 ± 0.0007, p = .005). Current smoking and former smoking were associated with lower serum CTX concentrations (−8%, p = .02, and −11%, p = .02) but not with serum P1NP. The number of pack-years also was inversely related to CTX (β = −0.003 ± 0.0007, p < .001) but not to P1NP (p = .35). Parathyroid hormone (PTH) levels were not related to current smoking or number of pack-years (data not shown).

Smoking in relation to gonadal steroids as determinants of bone parameters

Gonadal steroids and physical activity as determinants of bone geometry and vBMD have been described previously in this cohort.19 Free estradiol was the major sex steroid positively associated with cortical vBMD and cortical thickness and negatively associated with endosteal circumference at the radius.19 Educational level was positively associated with cortical thickness at the tibia (β = 0.003 ± 0.001, p = .04), whereas alcohol and calcium intake were not consistently associated with bone parameters.

A significant interaction term was found between current smoking and free estradiol in predicting cortical thickness, cortical bone area, and endosteal circumference whereby the negative effects of smoking on bone size were attenuated with higher estradiol levels (Table 3). Adjusting these models for possible confounding effects of education level, calcium intake, physical activity, and alcohol use did not alter these associations for cortical thickness, endosteal circumference, or trabecular density. No major interaction with free estradiol was found using former smoking in the statistical models or with bone parameters at the radius. When analyzing the association between gonadal steroids and bone parameters separately in smokers and never smokers (Table 4), strong positive associations were found between free estradiol and cortical thickness and negative associations with endosteal circumference in smokers, whereas in never smokers no significant association was found (see Table 4). Free testosterone demonstrated a modest interaction with smoking in the mixed-effects models predicting cortical thickness and cortical bone area (see Table 3). In the group of current smokers, free testosterone was associated with cortical thickness and cortical area but not with endosteal circumference (see Table 4). In nonsmokers, no associations between testosterone and the mentioned bone parameters were observed.

Table 3. Cross-Sectional Associations Between Lifestyle Factors, Hormonal Concentrations, and Trabecular Bone Parameters at the Radius and Cortical Bone Parameters at the Tibia

	Radius		Tibia
	Trabecular bone area	Trabecular vBMD	Cortical bone area	Cortical vBMD	Cortical thickness	Endosteal circumference
Data presented as estimates of effect size (β ± SEM). * p ≤ .001; ** p ≤ 0.01; *** p ≤ .05. Bone parameters were log-transformed prior to analysis. Estimates for smoking refer to current smoking versus never smoking. Former smoking exhibited no interaction with free estradiol in the statistical models. Free testosterone was analyzed separately from free estradiol using the same fixed effects to avoid collinearity.
Current smoking	0.10 ± 0.05***	−0.16 ± 0.06**	−0.008 ± 0.04***	−0.080 ± 0.008	−0.15 ± 0.05**	0.10 ± 0.04**
Age	0.004 ± 0.001*	−0.003 ± 0.001***	0.0003 ± 0.001	−0.0001 ± 0.0002	−0.0003 ± 0.001	0.001 ± 0.001
Height	0.007 ± 0.001*	−0.006 ± 0.001*	0.006 ± 0.001*	−0.00003 ± 0.0002	0.002 ± 0.001***	0.004 ± 0.0006*
Weight	0.0003 ± 0.0005	0.002 ± 0.001*	0.003 ± 0.0004*	−0.0004 ± 0.0001*	0.001 ± 0.0005**	0.002 ± 0.0004*
Free testosterone	−0.002 ± 0.002	0.003 ± 0.003	−0.002 ± 0.002	−0.0001 ± 0.0003	−0.002 ± 0.002	0.001 ± 0.002
Free estradiol	0.015 ± 0.059	0.05 ± 0.008	−0.019 ± 0.05	0.014 ± 0.009	0.017 ± 0.020	0.002 ± 0.04
Interaction terms
Smoking × free testosterone	0.0002 ± 0.003	0.006 ± 0.004	0.008 ± 0.003**	0.0006 ± 0.0006	0.008 ± 0.003**	−0.002 ± 0.002
Smoking × free estradiol	−0.19 ± 0.11	0.34 ± 0.15***	0.085 ± 0.043	0.018 ± 0.019	0.29 ± 0.11**	−0.19 ± 0.08***

Table 4. Associations Between Hormonal Concentrations and Cortical Bone Parameters at the Tibia in Both Current Smokers and Never Smokers

	Never smokers			Current smokers
	Cortical area	Cortical thickness	Endosteal circumference	Cortical area	Cortical thickness	Endosteal circumference
Data presented as estimates of effect size (β ± SEM). Free testosterone was analyzed separately from free estradiol using the same fixed effects to avoid collinearity.
Age	−0.0003 ± 0.001 p = .81	−0.001 ± 0.001 p = .43	0.001 ± 0.001 p = .28	0.001 ± 0.002 p = .43	0.0001 ± 0.002 p = .96	0.0015 ± 0.0015 p = .34
Height	0.006 ± 0.001 p <.001	0.002 ± 0.001 p = .14	0.005 ± 0.001 p < .001	0.006 ± 0.002 p = .001	0.002 ± 0.002 p = .18	0.004 ± 0.001 p = .013
Weight	0.004 ± 0.0006 p < .001	0.002 ± 0.0006 p = .003	0.002 ± 0.0005 p < .001	0.002 ± 0.001 p = .04	0.001 ± 0.001 p = .46	0.001 ± 0.001 p = .10
Free testosterone	−0.001 ± 0.002 p = .39	−0.002 ± 0.002 p = .24	0.001 ± 0.002 p = .38	0.006 ± 0.003 p = .02	0.007 ± 0.003 p = .02	−0.002 ± 0.002 p = .42
Free estradiol	−0.017 ± 0.06 p = .75	−0.009 ± 0.06 p = .88	−0.011 ± 0.04 p = .81	0.17 ± 0.09 p = .08	0.30 ± 0.11 p = .009	−0.22 ± 0.09 p = .016

Fracture prevalence in relation to physical activity and smoking

Self-reported fractures were more prevalent in smokers than in never smokers. Table 5 presents the odds ratios for fracture prevalence associated with smoking behavior categorized in current or former smoking and early or late smoking. After adjustment for confounding effects such as age, height, weight, calcium intake, alcohol intake, and educational level, odds ratios (ORs) were not altered dramatically, and increased ORs for prevalent fractures were systematically observed for early smokers when considering all fractures (OR 2.13, 95% confidence interval [CI] 1.42–3.20), childhood fractures excluded (OR 1.85, 95% CI 1.11–3.09), or when considering only fractures after the onset of smoking (OR 1.86, 95% CI 1.10–3.16). Starting to smoke at an age later than 16 years was not associated with higher fracture prevalence after adjustment for confounders considering only fractures after onset of smoking or when excluding childhood fractures (see Table 5).

Table 5. Prevalence Odds Ratios of Smoking, Predicting Self-Reported Fractures in Normal Healthy Males, Adjusted for Age

	Prevalence odds ratio
	All fractures	Childhood fractures <15 years excluded	Fractures after onset of smokingb
a p ≤ .05, adjusted for age, height, weight, calcium intake, education level, and alcohol use; b Childhood fractures (<15 years of age) were excluded in never smokers.
Smoking (current vs. never)	2.13 (1.43–3.18)ap = .0002	2.09 (1.27–3.44)ap = .004	1.81 (1.06–3.07) p = .03
Smoking (former vs. never)	1.90 (1.25–2.93)ap = .003	1.70 (1.00–2.90) p = .05	1.28 (0.72–2.27) p = .39
Smoking (early vs. never)	2.22 (1.50–3.27)ap = .0001	2.00 (1.23–3.26)ap = .005	1.96 (1.18–3.24)ap = .009
Smoking (late vs. never)	1.75 (1.10–2.77)ap = .02	1.84 (1.06–3.22) p = .03	1.11 (0.59–2.10) p = .75

No association between (free) estradiol or (free) testosterone and fracture risk was observed, and no interaction between estradiol and smoking was observed predicting fracture prevalence.

Discussion

In this study of healthy male siblings, we demonstrated unfavorable effects of smoking on bone density and geometry in young men and increased fracture prevalence odds ratios in smokers. Especially in men who started smoking at an early age, we found lower bone density and a thinner bone cortex, suggestive for a disturbed acquisition of bone during puberty. In subjects who started to smoke after puberty, the effects of smoking were less pronounced.

Previously, the effect of smoking on bone has been investigated mainly in older populations with conditions of bone loss.20–22 Our observation of lower bone mass in early smokers provides evidence that smoking can interfere with the acquisition of peak bone mass and that timing of smoking in relation to pubertal bone expansion is important. Previous findings in the Gothenberg Osteoporosis and Obesity Determinants (GOOD) study cohort (healthy men 18 to 19 years of age) describing a lower cortical thickness in smokers3 are in agreement with our data, although no mechanism was put forward in the GOOD study. In smokers, we observed a wider endosteal cavity, a thinner bone cortex, and a lower trabecular density, which are all bone parameters determined by estrogens.

Supporting a role for estradiol in the acquisition of peak bone mass in men, our group previously described a deficient acquisition of lumbar spine bone mass, together with lower estradiol concentrations, in men with idiopathic osteoporosis but also in their sons.13, 14 Moreover, observations in an aromatase-deficient young man demonstrated that bone mass increased under estrogen treatment and that estrogens play an important role in the acquisition of peak bone mass in men.23

Since we found an interaction between smoking and estradiol in the statistical models predicting bone parameters, we hypothesize that early smoking can interfere with the buildup of peak bone mass through interaction with the gonadal steroids, especially estradiol.

In women, several studies indicate that smoking influences ovarian function, endogenous and exogenous estrogen action, and metabolism.24–29 Estradiol concentrations were lower in smoking premenopausal women than in nonsmokers. Animal data demonstrated lower ovarian synthesis of estradiol and inhibition of ovulation on exposure to nicotine.25 In early postmenopausal women, smoking reduced the effect of low-dose estrogen supplementation on BMD, but the effect was maintained with higher estrogen treatment (2 mg estradiol).27–29 Although the exact mechanism cannot be derived from our observational study, several possible explanations can be formulated. First, an increased metabolism of estradiol in smokers, leading to a lower efficiency of hormonal replacement therapy in women or estrogen action, could occur. Indeed, in both males and females, increased estradiol 2-hydroxylation was found, producing the peripherally inactive catechol estrogens 2-hydroxyestrone and 2-methoxyestrone,30, 31 which could lead to lower circulating estradiol concentrations. Second, nicotine and its metabolites have been described to inhibit aromatase activity in vitro, which could lead to decreased estrogen synthesis.32 Direct interaction between toxic components in cigarette smoke and the estrogen receptor also could occur, which could weaken estrogen action.

In our cohort, we did not observe lower estradiol concentration in smokers, although significantly higher testosterone concentrations were observed, in agreement with previous findings in our laboratory,33and the lower estradiol:testosterone ratios support aromatase inhibition or 2-hydroxylation degradation of estradiol in smokers. Moreover, even in young smokers (12 to 14 years of age), similar hormonal disturbances have been found.34

Third, an obvious explanation, proposed in previous publications,35, 36 is the lower body mass index (BMI) in smokers, which could explain the lower bone mass. However, height, weight, and BMI were comparable between smokers and nonsmokers in our study population, supporting direct effects of smoking on bone rather than indirect effects of smoking through body composition. PTH was not different between smokers and nonsmokers, in contrast to previous reports in postmenopausal women.37

Finally, a direct effect of smoking on bone has been proposed in previous reports, although the molecular mechanism by which smoking affects bone metabolism is unclear, and incongruous effects of smoking on osteoblasts and osteoclasts have been described. Our data support a decreased bone turnover in smokers because bone resorption markers are inversely associated with smoking. In line with our findings, in vitro experiments demonstrated that nicotine inhibited both the formation of tartrate-resistant acid phosphatase (TRAP)–positive multinucleated cells (MNCs) and the resorption of bone by osteoclast-like cells.38 Taken together, we provide arguments that the low bone mass in smokers is probably due to deficient bone acquisition and less related to actual or persistent bone loss. In older men, however, smoking was associated with higher bone markers,4 which is at variance with our findings in young men. In men older than age 60, bone mass declines with advancing age, and in general, bone loss is present, which could explain this difference.

On the other hand, estrogens have been described to enhance nicotine metabolism by induction of the nicotine-metabolizing enzyme CYP2A6 in an estrogen receptor α–dependent manner. In this view, a high estrogen status could accelerate nicotine breakdown, limiting the negative effects of smoking.39, 40 Indeed, in current smokers, we found protective effects of higher estradiol concentrations toward the negative effects of smoking on bone parameters.

Moreover, in mice, estrogen deficiency was found to increase age-related bone loss by decreasing the defense mechanisms against oxidative stress.41 In all organ systems, smoking markedly increases oxidative stress42 and could be related to increased bone loss. In line with the observations in mice, we demonstrate protective effects of estrogens toward the unfavorable effects of smoking on bone parameters, which could be mediated by increased oxidative stress.

In our cohort, smoking was associated with prevalent fractures, occurring after childhood, with higher odds ratios observed in early smokers than in late smokers. In our analyses, we have excluded childhood fractures (<15 years of age) because these occur before the onset of smoking. The incidence of fractures according to age in our cohort (Fig. 1) corresponds to previous reports on fracture incidence in Britain,43 with peak incidences at 14 to 15 years of age.

However, smoking is a part of a set of (unhealthy) lifestyle factors that all could contribute to lower bone mass, and it is difficult to ascribe risk profiles to individual risk factors because smoking is associated with a lower degree of education, and alcohol use, and smokers are less conscientious about healthy living. In our cohort, this clustering of risk factors is also present because smoking was found to be associated with a lower level of education, higher manual labor, less participation in sports, and higher alcohol use. Taking these factors into account, smoking remained associated with increased fracture risk and smaller cortical thickness.

Two limitations of our study should be discussed. First, fractures were self-reported in our cohort and were not verified by systematic radiographs. Under- and overreporting can occur owing to poor recall.44 However, our fracture incidence corresponds to the published data,43 and smokers and nonsmokers probably will underreport to the same degree and therefore would not have influenced our findings. Second, the determination of estradiol by immunoassay has a lower precision at low estradiol concentrations. Measurement of estradiol by mass spectrometry can improve precision but is limited by the availability of laboratories that offer high-throughput mass spectrometry. In this study, we modified our protocol using a double amount of serum to increase precision at low concentrations of estradiol.19 Moreover, the associations between bone parameters and estradiol, assessed by immunoassay or by mass spectrometry, are comparable.45

In conclusion, we demonstrated unfavorable effects of smoking on areal BMD and bone geometry in healthy young men at age of peak bone mass, with increased fracture prevalence. Moreover, we observed an interaction between smoking and estradiol concentrations, providing arguments for a disturbed acquisition of peak bone mass during puberty by smoking, possibly mediated by an impaired sex steroid action.

Disclosures

The authors state that they have no conflicts of interest.

Acknowledgements

Both YT and BL contributed equally to this article. YT is a postdoctoral fellow and GV holds a PhD-fellowship of the Research Foundation, Flanders (FWO). This study was supported by grants from the Fund for Scientific Research, Flanders (FWO-Vlaanderen Grant Nos. G.0332.02 and G.0662.07) and was made possible through the EEC-supported Network in Europe on Male Osteoporosis (NEMO), QLK6-CT-2002-00491.