Introduction

The anti-inflammatory, immunosuppressive, and antistress actions of glucocorticoids (GCs) play a protective role in vivo. Between 2.7 and 4.6 % of postmenopausal women are reported as currently taking oral GCs [1]. One of the principal complications of long-term GC use is a profound alteration of bone metabolism. Rapid bone loss is most marked on endocortical surfaces [2]. Fracture risk increases already during the first 6 months of therapy and is positively related to daily dose [3], emphasizing the medical need to understand the biology of GC-induced bone loss so that clinicians can effectively prevent and treat this disease [4]. Supraphysiological concentrations of GC induce an early upregulation of genes associated with osteoclast cytoskeletal reorganization and organic bone matrix degradation [5] and stimulate bone resorption by promoting osteoclast survival and activity [6, 7]. GCs also inhibit the Wnt/β-catenin signaling pathway, enhance bone marrow stromal cell development toward the adipocyte lineage rather than toward the osteoblast lineage, and rapidly and significantly impair proliferation and differentiation, survival, and function of osteoblasts [5, 8, 9]. GCs also induce the loss of osteocytes by apoptosis and by autophagy [10, 11]. Osteocytes produce negative regulators of the Wnt/b-catenin pathway such as Dickkopf-1 (Dkk-1) and sclerostin, the key components regulating the number and activity of bone-forming osteoblasts [12] as well as osteoprotegerin (OPG) and the receptor activator of nuclear factor kappaB ligand (RANKL), the key components involved in osteoclast differentiation [13]. These changes favor reduction in new bone formation, a decrease in bone mass, and impaired quality of bone [4, 14, 15].

In clinical practice, high-dose (15 mg/kg daily for 10 days) iv methylprednisolone caused an immediate and persistent decrease in the serum marker of type 1 collagen synthesis (amino-terminal propeptide of type I procollagen, PINP) and osteocalcin (OC) in 13 patients suffering from multiple sclerosis [16] and in nine asthmatics treated with high iv dosages of betamethasone (0.65 mg/h) [17]. Serum concentrations of PINP were also significantly reduced after 2 weeks’ exposure to a low dose of prednisone (5 mg daily) in healthy postmenopausal women [18]. Bone-specific alkaline phosphatase, another marker of osteoblast function, did not change significantly during administration of higher doses of GCs [16, 18]. Effects of GCs on bone resorption markers are dose-dependent. High-dose iv GCs cause a rapid and transient increase of serum marker of type-1 collagen degradation (type 1 collagen cross-linked C-telopeptide, βCTX) [16, 18]. The above studies, however, did not evaluate the acute-phase reactants to assess a possible association between reduced inflammation and changes in the bone turnover markers after GC administration. To further assess the effects of oral GCs in doses used in clinical practice on osteoclast, osteoblast, and osteocyte function, we measured serum concentrations of the proven markers of bone remodeling, such as βCTX, PINP, and OC [19], simultaneously with the other possible mediators of the effects of GCs on bone, including soluble RANKL (sRANKL); its decoy receptor OPG [20]; two inhibitors of the Wnt/β-catenin signaling pathway, sclerostin and Dkk-1 [12]; and the high-sensitivity C-reactive protein (hsCRP).

Subjects and Methods

Study Design

We prospectively enrolled adult patients suffering from various medical pathologies requiring systemic steroid therapy who were never before treated with GCs and in whom GC treatment was currently initiated. All participants gave their written informed consent before enrollment. The study protocol and informed consent documents were prepared according to the Declaration of Helsinki and approved by the local ethical review board.

Before inclusion into the study, patients underwent a clinical examination (a full clinical history including details of comorbidity, detailed personal history of rheumatic disease, fracture history, alcohol intake, smoking, height loss, family history of osteoporosis, and hip fracture) and physical and laboratory examination.

One day before the initiation of GC treatment and each day before sampling, patients were asked to fast from 8:00 p.m. overnight. They were instructed to maintain normal hydration. At baseline and then after 24, 48, and 96 h, venous blood samples were collected for the assessment of biochemical markers. Immediately after sampling, the blood was centrifuged and the serum directly stored at −70 °C.

End points were changes in serum markers of osteoblast, osteocyte, and osteoclast activity during the first 96 h after initiation of GC therapy.

Patients

Included were 17 patients, 3 premenopausal women (age range 19–36 years, body mass index [BMI] 18.1–24.8 kg/m2), 10 postmenopausal women (age range 19–87 years, BMI 19.8–31.9 kg/m2), and 3 men (age range 53–63 years, BMI range 23.6–37.3 kg/m2) indicated for treatment with GCs. The indications for treatment with GCs were as follows: rheumatoid arthritis (n = 5), polymyalgia rheumatica (n = 3), systemic lupus erythematosus (n = 2), Churg–Strauss syndrome (n = 1), polymyositis (n = 3), Takayasu arteritis (n = 1), antisynthetase syndrome (n = 1), and pyrophosphate arthropathy (n = 1). Exclusion criteria were Expanded Disability Status Scale greater than 5.5, history of diseases affecting bone, prolonged immobilization (more than 3 weeks), and/or treatment with drugs known to influence bone metabolism, including GCs, any time.

At baseline, three patients were treated with hydrochloroquine, three with methotrexate, one with cyclosporine, and one with cyclophosphamide at recruitment. None of these patients was on biological therapy during the study. None of these patients was bedridden. Patients received a mean prednisolone equivalent dose of 23.1 ± 12.7 mg/day (range 10–50). Additional calcium and vitamin D supplements were provided on the third day after completion of blood study sampling. For ethical reasons it was not possible to postpone initiation of treatment by 4 days in order to perform the control tests of baseline reproducibility of the markers enabling each patient to be his or her own control. Therefore, the control group consisted of nine healthy adult volunteers (six women, three men, mean age 38.9 ± 11.3 years) who participated in the study after their written informed consent was obtained.

Biochemical Analysis

Fasting morning blood samples were obtained by antecubital venipuncture at baseline and before GC administration each day during treatment.

An enzyme-linked immunosorbent assay (ELISA) was used according to the manufacturer’s instructions, and the results were read and calculated by the ELISA reader Tecan Sunrise (Schoeller Instruments, Prague, Czech Republic). The studied parameters included serum concentrations of sRANKL (Ampli-sRANKL ELISA; Biomedica Medizinprodukte, Vienna, Austria; intra-assay imprecision coefficient of variation (CV) 9 %, interassay imprecision CV 3 %, detection limit 0.02 pmol/L), OPG (Osteoprotegerin ELISA, Biomedica Medizinprodukte; intra-assay imprecision CV 4 %, interassay imprecision 8 %, detection limit 0.14 pmol/L), Dickkopf-1 (Dkk-1 ELISA, Biomedica Medizinprodukte; intra-assay imprecision CV 8 %, interassay imprecision CV 12 %, detection limit 0.38 pmol/L), and human sclerostin (Human SOST ELISA; USCN Life Science, Brussels, Belgium; detection range 0.312–20 μg/L, detection limit <0.081 μg/L).

Serum OC, PINP, βCTX, intact parathyroid hormone (iPTH), and 25-hydroxyvitamin D [25(OH)D] were determined using electrochemiluminescence-based immunoanalysis (Cobas Analyzer; Roche Diagnostics, Mannheim, Germany). hsCRP was assessed using the immunoturbidimetric test (AU 400; Beckman Coulter, Prague, Czech Republic). The within-run imprecision of βCTX was below 7 % for samples between 200 and 500 ng/L and below 10 % for very low βCTX concentration samples. The within-run imprecision of iPTH was 5.4 % at serum concentration 3.2 pmol/L and 4.0 % at 6.6 pmol/L. The within-run imprecision of OC was <5 % at concentrations of 11–40 μg/L. The within-run imprecision of PINP was below 5 % at concentrations of 20–90 μg/L. The within-run imprecision of hsCRP was 2 %. The serum concentration of calcium was determined using the colorimetric color test Calcium Arzenazo III on the AU 400 analyzer.

The above serum markers, hsCRP, calcium, and phosphorus, were measured at baseline and after 24, 48, and 96 h of GC treatment. Serum 25(OH)D was measured at baseline. Serum specimens collected from all sampling days were analyzed simultaneously at the end of the study.

Bone Densitometry

Bone mineral density (BMD) was measured using dual-energy X-ray absorptiometry by the bone densitometer GE Prodigy (GE Healthcare, Madison, WI; software 12.10.113) at the lumbar spine, proximal femur, and femoral neck. Short-term in vivo precision errors for the lumbar spine, total femur, and femoral neck were 0.7, 0.9, and 1.8 %, respectively. BMD was expressed as a T score.

Statistical Methods

Statistical analysis of data was performed using SigmaPlot 10 software package (Systat Software, Erkrath, Germany). Relationships between variables (absolute and percent change from baseline) were explored using Pearson product moment correlation coefficients (r and p values reported). Differences in percentage change between the times were assessed by analysis of variance. Where appropriate, the Tukey test was used for comparisons. Results were considered statistically significant at the level of 0.05.

Results

Baseline characteristics of patients under study are given in Table 1. At baseline, BMD at the lumbar spine or proximal femur was in the range of osteoporosis (≤−2.5 T score) in one patient and normal (>−1 T score) in five patients. Serum 25(OH)D levels were in the deficient range (<25 nmol/L) in 9 of 17 patients and in the insufficient range (25–50 nmol/L) in 4 of 17 patients; serum iPTH concentrations were in the normal range in 12 of 17 patients; in five they were 7.2–17.3 pmol/L, with normal serum calcium. Serum concentrations of hsCRP >5 mg/L were observed in 13 of 17 patients.

Table 1 Baseline characteristics of the population under study
Full size table

At baseline, no significant correlations were observed between age, BMI, iPTH, 25(OH)D, hsCRP, and the other biochemical indices, except for significant positive correlations observed between serum PINP and OC (r = 0.82, p < 0.001), OPG and OC (r = 0.83, p < 0.001), and OPG and PINP (r = 0.62, p = 0.008) and a significant negative correlation between PINP and hsCRP (r = −0.51, p = 0.036).

During 96 h of GC administration, a significant reduction of serum OC, PINP, OPG, and sclerostin was observed (Fig. 1). Compared with baseline, after 24, 48, and 96 h, hsCRP was reduced by 23.2 ± 21.3 % (nonsignificant), 44.6 ± 28.6 % (p < 0.05), and 66.2 ± 21.7 % (p < 0.05), respectively. During 96 h of GC administration, serum βCTX showed a significant percentual increase (Fig. 1), while the increase in Dkk-1 did not reach statistical significance. A significant positive correlation was found between serum concentrations of Dkk-1 and βCTX (Fig. 2) after 96 h of treatment with GCs. This correlation was not significantly influenced by baseline iPTH, 25(OH)D, and hsCRP serum concentrations (Table 2). Also, the significant positive correlation between serum OPG and PINP was not significantly influenced by baseline iPTH, 25(OH)D, and hsCRP serum concentrations. No significant correlations were observed between age or sex and the biochemical indices studied. The increase in serum sRANKL (by 57 %) and sRANKL/OPG (by 77 %) did not reach statistical significance.

Fig. 1
figure 1

Change from baseline in serum markers in 17 patients treated with glucocorticoids (mean ± SE). *p < 0.05 versus baseline (ANOVA on ranks). Dotted lines indicate controls. a βCTX (filled circles) and PINP (empty circles). b Dkk-1 (filled circles) and sclerostin (empty circles). c Osteoprotegerin (filled circles) and osteocalcin (empty circles)

Full size image
Fig. 2
figure 2

Correlation between serum concentrations of Dkk-1 and βCTX after 96 h of treatment with glucocorticoids (r = 0.73, n = 17, p = 0.001)

Full size image
Table 2 Pearson product moment correlation between baseline serum concentrations of 25(OH)D, iPTH, and hsCRP and changes from baseline to 96 h in serum biochemical indices in 17 patients treated with glucocorticoids (r and p values)
Full size table

A significant correlation was observed between change in hsCRP and change in PINP and between change in hsCRP and change in sRANKL/OPG ratio (Fig. 3). Daily dose of GCs was borderline inversely correlated with change from baseline in serum PINP (r = −0.482, p = 0.05) but not with the other indices.

Fig. 3
figure 3

Serum concentrations of hsCRP, PINP, and RANKL/OPG ratio (median values and 75 % CI) at baseline (empty circle), 24 h (filled square), 48 h (empty triangle), and 96 h (filled triangle). Change in hsCRP versus change in PINP, r = 0.68, p < 0.001. Change in hsCRP versus change in RANKL/OPG ratio, r = 0.25, p = 0.042

Full size image

Discussion

Our main objective in the current study was to understand whether the acute changes in serum biochemical markers of bone turnover are associated with changes in mediators of the effects of GCs on osteoblasts and osteocytes. We studied a heterogeneous and small group of adults suffering from various chronic, low-grade systemic inflammations requiring systemic steroid therapy. The subjects received a medium dose of oral GCs for 4 days. As expected, treatment with GCs resulted in a fast and significant drop in serum hsCRP, an acute-phase reactant. Serum hsCRP level has been reported to be a significant predictor of low-trauma fracture [21, 22], and circulating levels of inflammatory markers correlate with bone resorption markers and predict change in BMD [23, 24]. However, other studies did not find a significant correlation between hsCRP and serum C-telopeptide as a marker of bone resorption [22, 25]. In the present study, we also did not find an association between hsCRP and serum CTX at baseline. However, we observed a significant negative correlation between hsCRP and PINP at baseline, indicating reduced osteoblastic bone formation in patients with chronic inflammation. This is in agreement with histomorphometry of bone biopsy specimens indicating reduced bone formation in non-steroid-treated patients with rheumatoid arthritis [26] and with a reduction in the markers of bone formation in juvenile chronic arthritis [27].

In agreement with previous studies in healthy subjects [16, 17, 28, 29] and in patients suffering from various medical pathologies requiring steroid therapy [16, 17, 28, 29], we observed a significant drop in serum OC and PINP concentrations after starting treatment with GCs. Gene expression of both OC and procollagen-α1(I) is directly downregulated by GCs [30]. In this study, a significant reduction in serum OPG is in good agreement with inhibition by dexamethasone of OPG in human osteoblastic lineage cells [31]. OPG functions as a decoy receptor that is able to neutralize both the cell-bound and soluble forms of RANKL. In mice, the suppressive effect of GCs on spinal BMD, cortical thickness, and strength was prevented by OPG [32].

Interestingly enough, a decrease in the serum markers of bone formation was accompanied by a significant reduction in serum sclerostin. Osteocytes, but no other cells of the osteoblastic lineage, express sclerostin, a Runx2-dependent product of the SOST gene. Sclerostin is a potent antagonist of bone morphogenetic proteins critical for osteoblastogenesis (BMP −2, −4, −5, −6, and −7) and appears to bind to lipoprotein-related receptors 5 and 6 (LRP5 and LRP6) to inhibit canonical Wnt/β-catenin signaling [33]. Locally, sclerostin acts as a negative regulator of late-osteoblast/preosteocyte differentiation, inhibits osteoblast proliferation, promotes osteoblast apoptosis, and suppresses mineralization of osteoblastic cells [34]. Loss of sclerostin in humans causes the high-bone mass disorders Van Buchem disease [35] and sclerosteosis [36], and administration of an antisclerostin antibody increases bone formation in mice treated with GCs [37]. It was suggested that transient reductions in sclerostin levels with daily injections of PTH could contribute to the antiapoptotic signaling triggered by this regimen [38]. Conversely, sclerostin stimulates apoptosis of cultured human osteoblastic cells [39], and transgenic mice overexpressing sclerostin exhibit low bone mass. In this study, the significant drop in serum sclerostin was associated with a decrease, rather than an increase, in markers of bone formation, reflecting a GC-induced apoptosis of osteocytes [6, 40]. This is in good agreement with decreased circulating sclerostin levels in patients with endogenous hypercortisolism, which increase during biochemical remission of the disease [41].

The modest increase in serum βCTX peaking at 24–48 h of GC treatment is consistent with previous reports [16, 19]. Simultaneously, we observed only a trend to increase in serum sRANKL, which is reminiscent of the poor and nonconsistent association between circulating OPG and RANKL with biochemical markers of bone turnover and BMD [42], although when measured in the bone marrow high correlations can be observed [43]. Also, lack of sensitivity of currently available assays may impair correlation analysis with conventional bone marker measurements. Although GCs stimulate RANKL synthesis and reduce osteoclast death by apoptosis [6], the ability of osteoclasts to resorb bone is impaired due to cytoskeletal abnormalities that prevent osteoclasts from anchoring efficiently bone and developing an adequate brush border. The response of osteoclasts to macrophage colony-stimulating factor, a cytokine produced by osteoblasts, is specifically impaired [7].

In this study, a significant positive correlation was observed between increase in serum βCTX and Dkk-1, another Wnt inhibitor. Although Dkk-1 disrupts the Wnt cascade by binding to Wnt coreceptors LRP5 and LRP6, resulting in the inhibition of osteoblast differentiation [44], it could lead to increased osteoclastic activity and overall bone turnover through enhanced osteoblast-dependent osteoclastogenesis by upregulating RANKL and/or downregulating OPG [45]. Expression of Dkk-1 in human osteoblasts and osteocytes is enhanced by GCs and may lead to enhanced osteoblast-dependent osteoclastogenesis [46], osteoblast apoptosis, and bone mass loss [47, 48].

Dickkopf-1 is highly expressed in osteocytes as well as in several other cell types. Despite this, serum concentrations of Dkk-1 are assumed to reflect expression of Dkk-1 in the bone microenvironment and predict the extent of some bone diseases [49, 50]. In a cohort study in postmenopausal women with established osteoporosis, serum Dkk-1 levels were significantly higher in osteoporotic women compared with controls [51].

Despite an initial significant negative correlation between hsCRP and PINP values, administration of GCs was associated with a significant drop in both serum hsCRP and PINP concentrations. The positive correlation between hsCRP and the sRANKL/OPG ratio after GC administration was driven mainly by the drop in serum OPG concentrations. Thus, in this study, the reduced inflammation was not associated with reduction in sRANKL and serum CTX but, rather, was associated with decreased serum concentrations of markers synthesized by osteoblasts and osteocytes (PINP, OC, OPG, and sclerostin). The results are in good agreement with negative effects of GCs on bone remodeling; however, they do not support a causal relationship between CRP, a sensitive marker of chronic, low-grade systemic inflammation, and the response of bone turnover markers to GC.

Several limitations of the study must be taken into account. First, due to the pilot observational design of the study and the small number of subjects, the conclusions must be interpreted with caution. The relationship between hsCRP and PINP in non-steroid-treated patients with diseases has not been studied previously and requires confirmation by a larger study. Long-term blood sampling and BMD measurements would be required to study changes and differences in the markers between different dosages of GCs. Second, it is unclear to what extent serum concentrations of sclerostin and RANKL correlate with local cytokine production or action within the bone microenvironment. In addition, we did not measure inflammatory cytokines, such as IL-1, IL-6, and TNF-α. Third, our group of adult patients suffering from various medical pathologies requiring systemic steroid therapy was heterogeneous in terms of age, menopausal status, inflammatory activity of the primary disease, and vitamin D serum concentrations. The latter reflects real clinical practice where initiation of GC therapy cannot be held off until vitamin D saturation. Three of the patients were treated with hydrochloroquine, a known inhibitor of 25-(OH)D-1α-hydroxylase which could result in low serum 1,25(OH)2D levels with consequent effects on osteoblasts. However, excluding these patients did not modify the results significantly (data not shown). The number of subjects in this study was too small to allow a search for differences among subgroups and perhaps to see an expected correlation between 25(OH)D levels and PTH. We acknowledge that, for ethical reasons, each patient could not be used as his or her own control. This cannot be substituted with the current control group. Therefore, the changes in the markers after administration of GCs were compared with the baseline values, and the control group was used to document the reproducibility of the measurements in healthy adults.

In conclusion, medium-dose, short-term oral GC regimens cause an immediate decrease of the biochemical markers of osteoblast and osteocyte activity (PINP, OC, OPG, and sclerostin) and a moderate increase of the biochemical marker of bone resorption (βCTX). Other randomized studies are warranted to test the markers and mediators of the effects of GCs on bone to identify high–fracture risk patients and to implement preventive therapy before rapid bone loss occurs.