- Top of page
- Materials and Methods
Background: There is a potential interface between osteoporosis and the chronic inflammation of inflammatory bowel disease (IBD), and the osteoprotegerin (OPG)/receptor for activated nuclear factor-κB (RANK)/RANK ligand (RANKL) signaling pathway may be an important mediator, although data are limited.
Methods: We conducted a population-based case-control seroassay study to look for alterations in serum OPG and soluble RANKL (sRANKL). The study population included IBD patients who were 18 to 50 years old with Crohn's disease (CD; n = 287) or ulcerative colitis (UC; n = 166), age-matched healthy controls (n = 368), and nonaffected siblings of IBD patients (n = 146). Serum OPG and sRANKL were measured by enzyme-linked immunoassay. Sex-specific reference ranges were derived from the healthy controls.
Results: Analysis of variance (ANOVA) confirmed significant group differences in women for mean serum OPG (P = 0.018). CD women had higher values of OPG than UC women (P = 0.028) or healthy controls (P = 0.045), whereas the other groups were similar. OPG levels were above the reference range in 13/173 (8%) of CD women, exceeding the expected proportion (P = 0.032). In contrast, no differences in OPG were seen in men between controls, CD, or UC. Estrogen use in women (P = 0.000002) and corticosteroid use in men (P = 0.026) were associated with higher OPG levels. In multivariate analysis, CD diagnosis (P = 0.031) and estrogen use (P = 0.000002) were independently associated with higher OPG levels. No group differences were seen in mean serum sRANKL measurements.
Conclusions: An OPG:sRANKL imbalance with OPG exceeding sRANKL should inhibit osteoclastogenesis and promote bone formation. CD is associated with increased fracture risk, and possibly, the paradoxically higher OPG is a counterregulatory response to factors such as inflammatory cytokines, promoting high bone turnover. Alternatively, elevated OPG in CD may reflect T-cell activation.
Crohn's disease (CD) is associated with a slight but statistically significant increased risk of fracture compared with the general population.1–3 While the elderly have the highest risk of fracturing,1–4 the increased risk is evident across all age groups.1 Approximately 15% of patients with CD seen at specialty clinics have osteoporosis as measured by dual energy x-ray absorptiometry testing.5 Hence, reconciling the high rates of osteoporosis at specialty clinics and the lower than predicted rates of fractures suggests that there is a minority of patients with CD who are at substantial risk of fracturing. Two major risk factors have been considered to play a role in osteoporosis in CD: systemic inflammation and the use of corticosteroids.5 While corticosteroid use has been shown by some but not all studies to be associated with osteoporosis,5 only recently has the use of corticosteroids been shown to be associated with fracture risk in CD in a population-based study.6 However, those with the highest degree of systemic inflammation are also the ones most likely to use corticosteroids, and so teasing out distinct effects of each factor is difficult.
Our understanding of the pathophysiology of normal bone homeostasis and the pathogenesis of osteoporosis and its integration with the immune response has been greatly advanced by the discovery of a receptor-ligand pathway identified on osteoblast and osteoclast precursors. Osteoblasts express a surface ligand [receptor-activator of nuclear factor kappa B (NF-κB) ligand (RANKL)], which can bind to osteoclast precursors [the receptor activator of NF-κB (RANK)] or an osteoblast derived soluble decoy receptor-osteoprotegerin (OPG).7 The binding of RANK to RANKL induces a signaling and gene expression cascade that results in differentiation and maturation of osteoclasts that can ultimately lead to osteoporosis. OPG blocks this interaction, thereby inhibiting osteoclast formation and possibly interfering with osteoporosis. It has been shown that agents that can enhance RANKL production are associated with osteoporosis, whereas agents that enhance OPG reduce osteoporosis.8 Corticosteroids, for instance, enhance RANKL and inhibit OPG. Simplistically, one would anticipate that states associated with osteoporosis might be more likely to have lower OPG and higher RANKL.
RANKL is also a regulator of T cell-dendritic cell interaction in the immune system and is a crucial factor in early lymphocyte development and lymph node organogenesis.9 The central importance of this system is seen in that RANKL gene-deficient mice, which are unable to support osteoclast differentiation, display severe osteopetrosis (even in the presence of bone resorbing factors such as vitamin D3, dexamethasone, and prostaglandin E), show no evidence of bone remodeling, and simultaneously lack all lymph nodes.9 Activated T cells can directly trigger osteoclastogenesis through RANKL, leading to bone loss, an effect that is blocked by OPG.9–11 Hence, this system may be critical in linking systemic or mucosal inflammation with altered bone metabolism and, ultimately, osteoporosis.
We aimed to develop a serum assay to measure each of OPG and soluble RANKL (sRANKL) and to test it in a population-based case control sample of CD, ulcerative colitis (UC), and healthy controls to determine if there were differences between the groups, stratified by sex.
- Top of page
- Materials and Methods
The group demographics and characteristics are summarized in Table 1. They were well matched for age, ethnicity, and current smoking. As expected, corticosteroid use was largely confined to the CD and UC subgroups. There was slight excess estrogen hormone use (predominantly in the form of oral contraceptives) by CD women, and slightly more current smoking by CD and UC women, but this was not statistically significant.
Table 1. Subject Demographics and Characteristics
| || || ||Healthy||Sibling|
|Women|| || || || |
|Age ± SD (yr)||38 ± 7||38 ± 7||40 ± 5||39 ± 9|
|White ethnicity (%)||97||92||96||88|
|Current sμoking (%)||27||26||18||21|
|Corticosteroid|| || || || |
|Estrogen horμone|| || || || |
|Men|| || || || |
|Age ± SD (yr)||35 ± 8||38 ± 8||40 ± 7||39 ± 8|
|White ethnicity (%)||95||94||93||92|
|Current sμoking (%)||25||21||23||22|
|Corticosteroid|| || || || |
Healthy women had significantly higher mean OPG values than healthy men (0.90 versus 0.80 pg/mL, respectively; P = 0.043). Therefore, women and men were analyzed separately. For consistency, sRANKL analyses were also performed according to sex, although no significant sex difference was detected (healthy women, 0.45 pg/mL; healthy men, 0.48 pg/mL; P > 0.2). OPG did not show any significant correlation with age in healthy control men or women (P > 0.2). There was a weak negative correlation between sRANKL and older age in women (r = −0.13, P = 0.035) but not in men (P > 0.2). When the analysis was extended to all men, no significant correlations with age were seen (P > 0.2). OPG and sRANKL were positively correlated in healthy women (r = 0.15, P = 0.012) but not in healthy men (P > 0.2). When the analysis was extended to all women, no significant correlation between OPG and sRANKL was seen (P > 0.2).
ANOVA confirmed significant group differences in women for mean serum OPG (P = 0.018; Table 2). Post hoc analysis showed that CD women had higher values of OPG than UC women (P = 0.028) or healthy control women (P = 0.045), whereas the healthy controls, sibling controls, and UC women were similar. The upper limit of normal for OPG was 1.87 pg/mL in women and 1.58 pg/mL in men, and the upper limit of normal for sRANKL was 1.65 pg/mL in women and 2.22 pg/mL in men. OPG levels were above the reference range in 13 of 173 (8%) CD women, and this exceeded the expected proportion of 2.5% (P = 0.032). In contrast, no differences were seen in serum OPG in men between healthy controls, sibling controls, CD, or UC. No group differences were seen in mean sRANKL measurements for women or men.
Table 2. Mean Serum OPG and sRANKL Levels
| ||CD||UC||Healthy Controls||Sibling Controls|
|Women|| || || || |
|OPG, μean (95% CI), pg/μL*||1.02 (0.95-1.11)||0.84 (0.76-0.94)||0.90 (0.84-0.95)||0.95 (0.85-1.06)|
|sRANKL, μean (95% CI), pg/μL||0.41 (0.38-0.45)||0.42 (0.38-0.48)||0.45 (0.42-0.48)||0.48 (0.42-0.54)|
|Men|| || || || |
|OPG, μean (95% CI), pg/μL||0.83 (0.75-0.91)||0.81 (0.73-0.91)||0.80 (0.72-0.88)||0.87 (0.76-0.99)|
|sRANKL, μean (95% CI), pg/μL||0.45 (0.4-0.51)||0.49 (0.42-0.57)||0.48 (0.42-0.55)||0.56 (0.47-0.68)|
Univariate analysis was undertaken to assess whether OPG was affected by factors other than diagnosis. Smoking and ethnicity did not affect levels of OPG (Table 3). Estrogen hormone use in women (P = 0.000002) and corticosteroid use in men (P = 0.026) were each associated with higher OPG levels. Two-way ANOVA was undertaken to test the significant univariate factors after adjusting for diagnosis. There was no evidence of a significant interaction between estrogen hormone use and diagnosis in women or corticosteroid use and diagnosis in men (interaction P > 0.2). In multivariate analysis, a diagnosis of CD (P = 0.031) and estrogen hormone use (P = 0.000002) were independently associated with higher OPG levels in women. Corticosteroid use in men remained statistically significant (P = 0.021) after adjusting for diagnosis. No significant factors affecting sRANKL were identified.
Table 3. Effect of Subject Characteristics (Other Than Diagnosis) on Serum OPG
| ||Unadjusted OPG, Mean (95% CI), pg/μL||Adjusted for Diagnosis OPG, Mean (95% CI), pg/μL|
| ||Factor Present||Factor Absent||Factor Present||Factor Absent|
|Woμen|| || || || |
|Current sμoking||0.99 (0.90-1.08)||0.91 (0.87-0.96)||0.98 (0.89-1.07)||0.91 (0.86-0.96)|
|Corticosteroid user||1.00(0.86-1.17)||0.92 (0.88-0.96)||0.97(0.83-1.13)||0.92 (0.87-0.96)|
|Estrogen horμone user (%)||1.19(1.05-135)*||0.89 (0.85-0.93)||1.18 (1.05-1.31)*||0.89 (0.85-0.93)|
|Men|| || || || |
|Current sμoking||0.85 (0.76-0.96)||0.82 (0.77-0.87)||0.86 (0.76-0.96)||0.82 (0.77-0.88)|
|Corticosteroid user||0.99 (0.85-1.16)†||0.80 (0.76-0.85)||1.02(0.84-l.23)t||0.81 (0.76-0.86)|
- Top of page
- Materials and Methods
There is precedence in the literature for assessing serum OPG and sRANKL levels in health and disease.15–21 In a healthy adult population, serum OPG has been reported to increase with age, with a sharp increase in women after age 60 and in men after age 70, but little effect of age in younger subjects <40 years old.16,17 By only evaluating subjects between 18 and 50 years of age, the effect of age on serum OPG and sRANKL was minimized. Similar to our findings, slightly higher OPG levels in women than men have been reported previously.18
Primary biliary cirrhosis and chronic hepatitis C, 2 diseases known to be associated with increased fracture rates,22 are associated with increased serum levels of OPG.19 Serum OPG was not associated with bone mineral density in a study of liver disease, and the authors suggested that the high serum OPG was related to active inflammation.19 Others found a direct correlation between serum OPG and erythrocyte sedimentation rate and the Larsen score (a disease activity score) in patients with rheumatoid arthritis, supporting the notion that serum OPG may be driven by systemic inflammation.20 Our observation of increased OPG in women with CD parallels these findings. This may at first seem counterintuitive because OPG interferes with osteoclast differentiation, conferring skeletal benefit, whereas these conditions are associated with bone loss. OPG may therefore be a protective host response that partially offsets the adverse skeletal effect created by the inflammatory state. This interpretation is supported by the lack of correlation between serum OPG and bone mineral density in the previously cited study of liver disease.19 Others have found a lack of correlation between serum OPG with bone mineral density.16,19,23,24
After cardiac transplantation, serum OPG progressively decreases.25 This is also a time when bone mineral density decreases rapidly, in part, related to the antirejection regimens that may include corticosteroids and cyclosporine. The fall in OPG may have a permissive effect contributing to acute bone resorption. This seems to contrast with chronic inflammatory states, such as CD, where serum OPG levels do not decrease (and may actually increase) and bone loss declines much more slowly.
Short-term corticosteroids given for active CD lead to transient reductions in OPG at 2 weeks, but by 12 weeks after therapy initiation when all subjects were free of corticosteroids, OPG levels returned to baseline.26 In another study, no correlation was evident between corticosteroids and serum OPG.27 In our study, corticosteroid use affected OPG levels in men only. The elevated serum levels among corticosteroid users might be considered surprising unless corticosteroid effects on serum levels differ from their effects on cellular levels (where these agents typically reduce OPG), or alternatively, the corticosteroid users may have had diminished bone mass, and the increased serum level might be seen as a response to diminished bone mass. The basis for the sex-specific effects of corticosteroids (in men) and CD (in women) is unclear and requires further study. In women, estrogen hormone use strongly affected OPG levels. Others have reported that OPG levels were significantly higher in healthy young women on oral contraceptives than in nonusers.28 In postmenopausal women, there is a significant but weak positive correlation between serum OPG and serum estradiol levels.24 OPG is increased in anorectic women and may be a compensatory response to the bone loss seen in this population.29 Nutritional factors could also be contributing to the higher OPG levels seen in women with CD in our study.
OPG can be produced by a variety of tissues and cell types other than skeletal osteoblasts. For example, OPG is also expressed by endothelial cells30 and the media of arteries in wild-type mice,31 suggesting a possible role in vascular biology. Collin-Osdoby et al32 showed that inflammatory cytokine activation of human microvascular endothelial cells caused a dramatic up-regulation of OPG and even more sustained RANKL expression. OPG can prolong endothelial cell survival by inhibiting apoptosis.33 Serum OPG is increased in patients with advanced coronary artery disease compared with subjects with normal coronary arteries.34 If CD is a vasculitis,35 elevations seen in serum OPG may reflect vascular release of OPG in response to inflammation rather than necessarily being released as a response to bone resorption.
OPG:sRANKL imbalance with OPG exceeding sRANKL should inhibit osteoclastogenesis and promote bone formation. CD is associated with increased fracture risk, and it is possible that the paradoxically higher OPG is a counterregulatory response to other factors (such as inflammatory cytokines) promoting high bone turnover. Unfortunately, we do not have direct evidence (from either bone biopsy or biochemical markers of bone turnover) that bone resorption is indeed increased. Compelling evidence for the importance of OPG in maintaining bone health during an intestinal inflammatory state comes from an animal model, which simultaneously suggested a role for OPG in intestinal inflammation.36 This study used the interleukin-2 (IL-2)-deficient mouse model of colitis, which is known to develop both osteopenia and colitis. Study animals had elevated levels of bone marrow mononuclear cell expression of sRANKL and OPG mRNA, as well as circulating sRANKL and OPG, compared with control littermates. Osteopenia was not evident in IL-2-deficient mice cross-bred to be T-cell deficient, and osteopenia could be induced in T-cell-deficient mice by adoptive transfer of T cells from IL-2-deficient mice. These data suggest that activated T cells are critical for mediating the osteopenia. Importantly, exogenous OPG administration reversed both the osteopenia and the colitis. The colitis was found to be abrogated by a specific reduction in colonic dendritic cells, whereas circulating inflammatory cytokines were unaffected by exogenous OPG. These data, therefore, show directly the importance of OPG in osteopenia and colitis in IL-2-deficient mice and the importance of activated T cells in mediating these conditions. This suggests that increased OPG in CD may be generated as much by intestinal inflammation as in response to osteopenia. Furthermore, this study provides direct experimental evidence for the human finding of elevated OPG levels in a setting of concurrent intestinal inflammation and osteopenia. More recently, Vidal et al37 showed, in humans, that OPG is constitutively produced by intestinal epithelial cells, is up-regulated by tumor necrosis factor-α, and may be important in mucosal immunoregulation and bone physiology. Exogenously administered OPG may even warrant clinical study as a novel therapeutic approach in CD.
In summary, we have found that female patients with CD have elevated levels of serum OPG. Whether this is a response to osteopenia, intestinal inflammation, nutritional factors, or a combination is unclear. Further study of the role of this protein in the immunopathogenesis of CD and bone disease that accompanies some patients with CD is necessary.