Abstract
The bone mass benefits of antiresorbers in postmenopausal osteoporosis are limited by the rapid coupling of decreasing bone resorption with bone formation. Wnt signaling is involved in this coupling process during treatment with bisphosphonates, whereas its role during treatment with the anti‐receptor activator of NF‐κB ligand (RANKL) antibody denosumab is unknown. The study population includes patients participating in a placebo‐controlled trial lasting 36 months: 19 women were on placebo and 24 on subcutaneous 60 mg denosumab every 6 months. All measured parameters (serum C‐terminal telopeptide of type I collagen [sCTX], serum bone alkaline phosphatase [bAP], Dickkopf‐1 [DKK1], and sclerostin) remained unchanged during the observation period in the placebo group. sCTX and bAP were significantly suppressed by denosumab treatment over the entire follow‐up. Denosumab treatment was associated with significant (p < 0.05) increases (28% to 32%) in serum sclerostin over the entire study follow‐up. Serum DKK1 significantly decreased within the first 6 months with a trend for further continuous decreases, which reached statistical significance (p < 0.05) versus placebo group from the 18th month onward. The changes in DKK1 were significantly and positively related with the changes in sCTX and bAP and negatively with hip bone mineral density (BMD) changes. The changes in sclerostin were significantly and negatively related only with those of bAP. The changes in bone turnover markers associated with denosumab treatment of postmenopausal osteoporosis is associated with significant increase in sclerostin similar to those seen after long‐term treatment with bisphosphonates and significant decrease in DKK1. This latter observation might explain the continuous increase over 5 years in BMD observed during treatment of postmenopausal osteoporosis with denosumab. © 2012 American Society for Bone and Mineral Research.
Introduction
Healthy bone undergoes continuous resorption by osteoclasts and replacement by osteoblasts. These processes are tightly balanced such that in the normal adult skeleton resorption and formation are in equilibrium. When resorption exceeds formation, bone loss occurs. The coupling of these two processes is the main limitation of the use of antiresorptive drugs for the treatment of osteoporosis. Suppression of bone resorption is followed within a few months by inhibition of bone formation, limiting the net gain in bone tissue to the so‐called “remodeling space.”1, 2 Although the mechanism underlying this coupling is unknown, it is hypothesized that osteoclasts either express “coupling factors” or cause them to be released from the bone matrix, which is known to contain large amounts of osteoblast growth factors.3
The identification of the Wnt/ß‐catenin signaling pathway as a major promoter of bone formation has led to considerable interest in potential crosstalk between this pathway and osteoclastic bone resorption.4 The Wnt pathway influences bone formation through effects on osteoblast number, maturation, and progenitor differentiation, and these actions are opposed by various intracellular and secreted factors.4 The secreted Wnt antagonists sclerostin and Dickkopf‐1 (DKK1) block Wnt signaling by binding to Wnt coreceptors, such as low‐density lipoprotein receptor‐related protein 5 (LRP5) and LRP6, and by inhibiting the canonical Wnt/β‐catenin signaling pathway.4, 5
It has been reported that changes in bone resorption markers associated with withdrawal of estrogen therapy are accompanied by increases in serum sclerostin values,6 which decrease as a consequence of estrogen exposure.6, 7 Teriparatide treatment is associated with short‐term decreases in serum sclerostin level8 but later increases in serum DKK1 values.9 We have recently shown that chronic treatment with bisphosphonates is accompanied by a gradual increase in circulating sclerostin level.9 All these results suggest that the changes in bone formation markers associated with bone active agents might result from adaptive changes in the Wnt system.
Our aim in this study was to evaluate serum levels of DKK1, sclerostin, and bone turnover markers in patients receiving long‐term treatment with another antiresorptive, denosumab, with a mechanism of action substantially different from that of bisphosphonates. Denosumab is a humanized monoclonal antibody directed against the receptor activator of NF‐κB ligand (RANKL).10 RANKL is mainly produced by osteoblasts and acts through the receptor activator of NF‐κB (RANK) found on osteoclasts and preosteoclasts.11 RANKL interacts with RANK and stimulates activation of osteoclasts, leading to augmented bone resorption.12 Denosumab administration in postmenopausal osteoporosis antagonizes RANKL activity, thus diminishing not only osteoclast activity as bisphosphonates but also their recruitment and maturation.13
Materials and Methods
The study population included the 43 women participating from our center in the FREEDOM study,1 a multicenter placebo‐controlled trial on the efficacy of denosumab (60 mg subcutaneously every 6 months) in postmenopausal osteoporosis. All patients received calcium (1 g/day) and vitamin D (800 IU/day) supplements. According to study design, serum samples were collected every 6 months just before the next denosumab dose, aliquoted, and stored at –70°C in our laboratory. The measurements included serum C‐terminal telopeptide of type I collagen (sCTX, a marker of bone resorption), serum bone alkaline phosphatase (bAP, a marker of bone formation), DKK1, and sclerostin from all patients at all available time points.
BAP and sCTX were measured by ELISA (IDS Ltd., Boldon, UK). Intra‐assay coefficients of variation in our laboratory were 2% for sCTX and 4% for bAP, whereas interassay variability were 8% for sCTX and 6% for bAP. Serum DKK1 and sclerostin were measured by ELISA (Biomedica Medizinprodukte GmbH & Co. KG, Wien, Austria) with a sensitivity of 0.38 pmol/L and 2.6 pmol/L and an intra‐assay coefficient of variation (CV) of 7% to 8% and 5%, respectively. Interassay variability was assessed on four separate occasions on four serum samples, and the CVs were 8.2% and 6.9% for DKK1 and sclerostin, respectively. Bone mineral density (BMD) was measured by dual‐energy X‐ray absorptiometry (DXA) at the spine and total hip by using a Hologic 4500 (Hologic Inc., Waltham, MA, USA).
Both the original study and this substudy were approved by the institutional review board of the Medical School of Verona. All women provided written informed consent for their participation in the study, which included the possibility to perform in stored serum samples any measurement related to bone metabolism or to safety issues.
All statistical analyses were performed per protocol by SPSS Version 17 (SPSS, Inc., Chicago, IL, USA). Analysis of variance (ANOVA) and then a two‐sided Student's t test were used to estimate the absolute differences between treatment groups. Significance was set after adjustment for multiple comparisons (Tukey's method). The relationships between variables of interest were defined using the Pearson correlation.
Results
The main characteristics of the study population are listed in Table 1. The denosumab and placebo groups were comparable for anthropometric, biochemical, and densitometric data at baseline. None of the analyzed patients discontinued treatment, and all participated in all follow‐up visits.
Demographics and Baseline Characteristics of Randomized Patients
| Characteristic | Control group (n = 19) | Denosumab group (n = 24) | p value |
| Age (years) | 72 ± 5 | 72 ± 4 | NS |
| Height (cm) | 154.95 ± 5.18 | 156.75 ± 5.28 | NS |
| Weight (Kg) | 59.53 ± 9.85 | 59.96 ± 8.27 | NS |
| BMI (Kg/m2) | 24.73 ± 3.53 | 24.37 ± 2.85 | NS |
| Spine BMD (mg/cm2) | 679 ± 58 | 688 ± 53 | NS |
| Spine BMD T‐score (SD) | −3.3 ± 0.4 | −3.3 ± 0.4 | NS |
| Hip BMD (mg/cm2) | 695 ± 99 | 710 ± 76 | NS |
| Hip BMD T‐score (SD) | −2.0 ± 0.8 | −1.9 ± 0.6 | NS |
| PTH (pg/mL) (normal range 23.2 ± 15.3) | 31.06 ± 13.17 | 25.29 ± 10.56 | NS |
| bAP (µg/L) (normal range 13.2 ± 4.7) | 12.02 ± 4.77 | 13.13 ± 4.12 | NS |
| CTX (µg/L) (normal range 0.75 ± 0.6) | 0.67 ± 0.35 | 0.73 ± 0.43 | NS |
| DKK1 (pmol/L) (normal range 45.7 ± 20.0a) | 100.47 ± 47.35 | 118.95 ± 69.96 | NS |
| Sclerostin (pmol/L) (normal range 29.9 ± 18.9) | 28.54 ± 11.37 | 31.53 ± 12.57 | NS |
| Characteristic | Control group (n = 19) | Denosumab group (n = 24) | p value |
| Age (years) | 72 ± 5 | 72 ± 4 | NS |
| Height (cm) | 154.95 ± 5.18 | 156.75 ± 5.28 | NS |
| Weight (Kg) | 59.53 ± 9.85 | 59.96 ± 8.27 | NS |
| BMI (Kg/m2) | 24.73 ± 3.53 | 24.37 ± 2.85 | NS |
| Spine BMD (mg/cm2) | 679 ± 58 | 688 ± 53 | NS |
| Spine BMD T‐score (SD) | −3.3 ± 0.4 | −3.3 ± 0.4 | NS |
| Hip BMD (mg/cm2) | 695 ± 99 | 710 ± 76 | NS |
| Hip BMD T‐score (SD) | −2.0 ± 0.8 | −1.9 ± 0.6 | NS |
| PTH (pg/mL) (normal range 23.2 ± 15.3) | 31.06 ± 13.17 | 25.29 ± 10.56 | NS |
| bAP (µg/L) (normal range 13.2 ± 4.7) | 12.02 ± 4.77 | 13.13 ± 4.12 | NS |
| CTX (µg/L) (normal range 0.75 ± 0.6) | 0.67 ± 0.35 | 0.73 ± 0.43 | NS |
| DKK1 (pmol/L) (normal range 45.7 ± 20.0a) | 100.47 ± 47.35 | 118.95 ± 69.96 | NS |
| Sclerostin (pmol/L) (normal range 29.9 ± 18.9) | 28.54 ± 11.37 | 31.53 ± 12.57 | NS |
NS = not significant; BMI = body mass index; SD = standard deviation.
The normal range was established from serum samples of 18 blood donors (BI‐20412 DKK1, Biomedica Medizinprodukte GmbH & Co. KG, Wien, Austria.)
Demographics and Baseline Characteristics of Randomized Patients
| Characteristic | Control group (n = 19) | Denosumab group (n = 24) | p value |
| Age (years) | 72 ± 5 | 72 ± 4 | NS |
| Height (cm) | 154.95 ± 5.18 | 156.75 ± 5.28 | NS |
| Weight (Kg) | 59.53 ± 9.85 | 59.96 ± 8.27 | NS |
| BMI (Kg/m2) | 24.73 ± 3.53 | 24.37 ± 2.85 | NS |
| Spine BMD (mg/cm2) | 679 ± 58 | 688 ± 53 | NS |
| Spine BMD T‐score (SD) | −3.3 ± 0.4 | −3.3 ± 0.4 | NS |
| Hip BMD (mg/cm2) | 695 ± 99 | 710 ± 76 | NS |
| Hip BMD T‐score (SD) | −2.0 ± 0.8 | −1.9 ± 0.6 | NS |
| PTH (pg/mL) (normal range 23.2 ± 15.3) | 31.06 ± 13.17 | 25.29 ± 10.56 | NS |
| bAP (µg/L) (normal range 13.2 ± 4.7) | 12.02 ± 4.77 | 13.13 ± 4.12 | NS |
| CTX (µg/L) (normal range 0.75 ± 0.6) | 0.67 ± 0.35 | 0.73 ± 0.43 | NS |
| DKK1 (pmol/L) (normal range 45.7 ± 20.0a) | 100.47 ± 47.35 | 118.95 ± 69.96 | NS |
| Sclerostin (pmol/L) (normal range 29.9 ± 18.9) | 28.54 ± 11.37 | 31.53 ± 12.57 | NS |
| Characteristic | Control group (n = 19) | Denosumab group (n = 24) | p value |
| Age (years) | 72 ± 5 | 72 ± 4 | NS |
| Height (cm) | 154.95 ± 5.18 | 156.75 ± 5.28 | NS |
| Weight (Kg) | 59.53 ± 9.85 | 59.96 ± 8.27 | NS |
| BMI (Kg/m2) | 24.73 ± 3.53 | 24.37 ± 2.85 | NS |
| Spine BMD (mg/cm2) | 679 ± 58 | 688 ± 53 | NS |
| Spine BMD T‐score (SD) | −3.3 ± 0.4 | −3.3 ± 0.4 | NS |
| Hip BMD (mg/cm2) | 695 ± 99 | 710 ± 76 | NS |
| Hip BMD T‐score (SD) | −2.0 ± 0.8 | −1.9 ± 0.6 | NS |
| PTH (pg/mL) (normal range 23.2 ± 15.3) | 31.06 ± 13.17 | 25.29 ± 10.56 | NS |
| bAP (µg/L) (normal range 13.2 ± 4.7) | 12.02 ± 4.77 | 13.13 ± 4.12 | NS |
| CTX (µg/L) (normal range 0.75 ± 0.6) | 0.67 ± 0.35 | 0.73 ± 0.43 | NS |
| DKK1 (pmol/L) (normal range 45.7 ± 20.0a) | 100.47 ± 47.35 | 118.95 ± 69.96 | NS |
| Sclerostin (pmol/L) (normal range 29.9 ± 18.9) | 28.54 ± 11.37 | 31.53 ± 12.57 | NS |
NS = not significant; BMI = body mass index; SD = standard deviation.
The normal range was established from serum samples of 18 blood donors (BI‐20412 DKK1, Biomedica Medizinprodukte GmbH & Co. KG, Wien, Austria.)
The BMD changes are shown in Fig. 1. In patients treated with denosumab, both spine and total hip BMD increased significantly versus both baseline and the placebo group. All measured serum markers (sCTX, bAP, DKK1, and sclerostin) remained unchanged during the observation period in the placebo group. Bone turnover markers (sCTX and bAP) were significantly suppressed by denosumab treatment over the entire follow‐up (Fig. 2). Denosumab treatment was associated with significant (p < 0.05) increases in serum sclerostin level within the first 6 months (+29%), and these were maintained over the entire study period (Fig. 3). In the denosumab group, serum DKK1 values significantly (p < 0.05) decreased versus baseline within the first 6 months with a trend for further continuous decreases, which also reached statistical significance versus the placebo group from the 18th month onward (Fig. 3).
Percent changes from baseline in spine and total hip BMD in the patients treated with denosumab (white squares) and in the control group (closed circles). #p < 0.001 from baseline; *p < 0.05 versus control group; **p < 0.001 versus control group.
Percent changes from baseline in serum CTX and bAP in the patients treated with denosumab (white squares) and in the control group (closed circles). #p < 0.001 versus baseline; *p < 0.001 versus control group.
Percent changes from baseline in DKK1 and sclerostin in the patients treated with denosumab (white squares) and in the control group (closed circles). #p < 0.05 versus baseline; *p < 0.05 versus control group.
The changes at all time points (both absolute and percent) versus baseline in DKK1 serum levels were significantly and positively related with the changes in serum levels of sCTX and bAP, whereas those of sclerostin were significantly and negatively related only with those of bAP (Table 2).
Correlation (r) Between Changes in DKK1 and Sclerostin With Those in Bone Turnover Markers (bAP and sCTX) and PTHa
| DKK1 (p value) | Sclerostin (p value) | |
| Percent changes | ||
| PTH | −0.037 (NS) | −0.127 (0.05) |
| bAP | 0.260 (<0.001) | −0.171 (0.003) |
| sCTX | 0.303 (<0.001) | −0.093 (NS) |
| Absolute changes | ||
| PTH | −0.076 (NS) | −0.117 (NS) |
| bAP | 0.359 (<0.001) | −0.207 (<0.001) |
| sCTX | 0.117 (0.06) | −0.141 (0.03) |
| DKK1 (p value) | Sclerostin (p value) | |
| Percent changes | ||
| PTH | −0.037 (NS) | −0.127 (0.05) |
| bAP | 0.260 (<0.001) | −0.171 (0.003) |
| sCTX | 0.303 (<0.001) | −0.093 (NS) |
| Absolute changes | ||
| PTH | −0.076 (NS) | −0.117 (NS) |
| bAP | 0.359 (<0.001) | −0.207 (<0.001) |
| sCTX | 0.117 (0.06) | −0.141 (0.03) |
NS = not significant.
Correlation (r) Between Changes in DKK1 and Sclerostin With Those in Bone Turnover Markers (bAP and sCTX) and PTHa
| DKK1 (p value) | Sclerostin (p value) | |
| Percent changes | ||
| PTH | −0.037 (NS) | −0.127 (0.05) |
| bAP | 0.260 (<0.001) | −0.171 (0.003) |
| sCTX | 0.303 (<0.001) | −0.093 (NS) |
| Absolute changes | ||
| PTH | −0.076 (NS) | −0.117 (NS) |
| bAP | 0.359 (<0.001) | −0.207 (<0.001) |
| sCTX | 0.117 (0.06) | −0.141 (0.03) |
| DKK1 (p value) | Sclerostin (p value) | |
| Percent changes | ||
| PTH | −0.037 (NS) | −0.127 (0.05) |
| bAP | 0.260 (<0.001) | −0.171 (0.003) |
| sCTX | 0.303 (<0.001) | −0.093 (NS) |
| Absolute changes | ||
| PTH | −0.076 (NS) | −0.117 (NS) |
| bAP | 0.359 (<0.001) | −0.207 (<0.001) |
| sCTX | 0.117 (0.06) | −0.141 (0.03) |
NS = not significant.
Corresponding changes in bone turnover markers were significantly correlated with each other, whereas changes in DKK1 and sclerostin were not related to each other either for absolute value or percent change (results not shown).
Absolute changes at all time points versus baseline in DKK1 levels, but not in sclerostin levels, were negatively associated with changes in total hip BMD occurring both in the entire study cohort (r = 0.28, p < 0.001) and in the denosumab‐treated patients (r = 0.19; p = 0.007) (results not shown).
Discussion
We have shown that the changes in markers of bone turnover that result from long‐term treatment with denosumab of postmenopausal osteoporosis are associated with significant increase in serum levels of sclerostin and significant decrease in serum DKK1 levels.
Both DKK1 and sclerostin interact with LRP5 and LRP6 and antagonizes the canonical Wnt pathway.4, 5, 14–16 Sclerostin is almost exclusively expressed in osteocytes, and its expression decreases in response to mechanical loading17 or exposure to parathyroid hormone (PTH),8, 18, 19 and this may locally relieve endogenous Wnt inhibition and promote bone formation.18
We have recently shown that serum sclerostin level increases after long‐term treatment of postmenopausal osteoporosis with bisphosphonates.9 This suggests that the decrease in bone formation observed after several months of treatment with bisphosphonates may result at least in part from overproduction of sclerostin. In this study, we have observed increases in sclerostin levels of the same magnitude to those seen after treatment with the bisphosphonate neridronate,9 and this might suggest that the two drugs share a common action on this aspect of the regulation of bone formation. These results do not imply that this signaling protein is always or the only coupling factor between resorption and formation because it was reported that circulating estrogen levels are inversely associated with sclerostin levels20 and sclerostin bone marrow expression21 in postmenopausal women. On the same line is the observation that both estrogen21 and raloxifene7 administration decreases circulating sclerostin. The explanation for these different effects on circulating sclerostin among antiresorbers remains obscure, even though it should be underlined that in the study by Modder and colleagues,21 the exposure to estrogen was limited to 2 weeks, a lag time that does not allow for coupling bone formation to the suppressed bone resorption. However, in the study by Chung and colleagues,7 the treatment with raloxifene lasted more than 18 months and it was associated with decreasing circulating sclerostin levels, whereas, at variance with our observation, they were unable to see significant changes after bisphosphonate treatment. Thus, according to available evidence, it seems that inhibition of bone turnover by estrogen is associated with decreasing sclerostin levels, and when this is obtained by using bisphosphonates or denosumab, serum sclerostin values tend to rise.
DKK1 expression is widespread in embryonic mice, whereas in adults it is almost exclusively confined to osteoblasts and maturing osteocytes.22 In some pathological conditions, DKK1 is also expressed by other cells. Thus, in patients with multiple myeloma DKK1 is expressed in plasma cells and its serum levels are positively correlated with the presence of bone lesions.23, 24
A role for increased DKK1 levels in bone loss was also demonstrated in patients with monoclonal gammopathy of undetermined significance (MGUS), a precursor condition of multiple myeloma in which patients have an increased risk of osteoporotic fractures.25
DKK1 was recently recognized as a key player in joint damage of rheumatoid arthritis.26 In ankylosing spondylitis, the bone formation–promoting factors functionally prevail, and this was at least partially attributed to decreased DKK1.27
We have recently observed that treatment with bisphosphonates is not associated with appreciable changes in DKK1 values,28 whereas we have shown here that denosumab therapy is associated with substantial and progressive decreases in DKK1 levels.
As far as we are aware, this is the first time that a decrease in circulating DKK1 is experimentally induced. The explanation for this observation and the differential effect with bisphosphonates is unclear. Intuitively, it is likely related to the different mechanism of action on osteoclasts. Bisphosphonates act primarily on mature osteoclasts and they increase number and size of the precursors,29 whereas denosumab also affects the recruitment and maturation of osteoclasts.30 Of some interest is the opposite correlation we found between the changes in DKK1 or sclerostin and bone turnover markers. The negative correlation between changes in sclerostin and bAP are consistent with the hypothesis that the increase in sclerostin during denosumab therapy is associated with the decrease in bone formation that typically follows the inhibition of bone resorption. On the other hand, the positive correlations between changes in DKK1 and both bone turnover markers might suggest that DKK1 production is related to bone turnover (the lower the bone turnover, the lower DKK1). This hypothesis is consistent with our recent observation that chronic treatment with teriparatide9 and primary hyperparathyroidism (Viapiana and colleagues, unpublished data, 2012) are associated with progressive increases in DKK1.
The clinical relevance of our observation on the changes in DKK1 seen with denosumab but not with bisphosphonates remains uncertain, but it might be put in perspective of the continuous increase over 5 years in hip BMD observed during treatment of postmenopausal osteoporosis with denosumab.31 This hypothesis is somewhat corroborated by the negative correlation we found between the changes in hip BMD and those in DKK1 levels, which might also suggest that the effects of DKK1 changes are dominant relative to sclerostin changes with denosumab therapy.
A working hypothesis might be that denosumab has a dual effect on the Wnt system: The observed increase in sclerostin levels tends to depress bone formation, but this is balanced by the progressive decrease in DKK1, which is likely to attenuate the negative effect on bone formation of chronic denosumab treatment.
Disclosures
All authors state that they have no conflicts of interest.
Acknowledgements
We thank Caterina Fraccarollo and Cristina Bosco for the ELISA assays.
Authors' roles: Study design: DG, SA, MR. Study conduct: All. Data collection: DG, OV, CD, MRP. Data analysis: DG, SA. Data interpretation: DG, SA, MR, OV. Drafting manuscript: SA, DG. Revising manuscript content: OV, LI, EF, DG. Approving final version of manuscript: All. SA takes responsibility for the integrity of the data analysis.
