Circulating semaphorin-4D and plexin-B1 levels in postmenopausal women with low bone mass: the 3-month effect of zoledronic acid, denosumab or teriparatide treatment

Abstract

Objective: The evaluation of circulating semaphorin-4D (sema4D) and plexin-B1 in postmenopausal women with low bone mass and the effect of antiresorptive or osteoanabolic treatment.

Methods: Serum samples were obtained from postmenopausal women with low bone mass at baseline and 3 months after zoledronic acid infusion (n = 30), denosumab injection (n = 30) or teriparatide initiation (n = 28) and from controls matched for age, age at menopause and body mass index (n = 30) at the same time points.

Main outcome measures: Circulating sema4D and plexin-B1.

Results: Circulating sema4D increased following denosumab (p = 0.026), whereas decreased following teriparatide (p = 0.013). Sema4D/plexin-B1 ratio increased following denosumab (p = 0.004). At baseline, sema4D and plexin-B1 levels were higher in patients pre-treated with bisphosphonates compared to naïve ones (p < 0.001 and p = 0.001, respectively). In bivariate correlations sema4D was inversely correlated with serum carboxyterminal telopeptide of type 1 collagen (r_s -0.282, p = 0.002), intact parathyroid hormone (r_s -0.388, p < 0.001) and 25(OH)D (r_s -0.316, p < 0.001), whereas there was a trend towards correlation with lumbar spine bone mineral density (r_s -0.191, p = 0.053).

Conclusions: Sema4D levels are independently associated with previous bisphosphonate treatment, intact parathyroid hormone and 25(OH)D levels. Denosumab and teriparatide seem to exert an opposite effect on circulating sema4D levels. Further studies are needed to evaluate whether sema4D mediates the coupling effect that occurs following both antiresorptive and osteoanabolic treatment.

Keywords:

• bisphosphonate
• denosumab
• osteoporosis
• plexin
• semaphorin
• teriparatide

1. Introduction

Skeletal renewal is achieved by bone remodeling, a process orchestrated by a continuous crosstalk between the bone cells involved. Although the exact mechanisms regulating this crosstalk are not fully elucidated, several pieces of the puzzle have been discovered during the past decade. Semaphorin-4D (sema4D), also known as CD100, is a molecule initially identified in immune cells, which is expressed by T cells and regulates immune cell activation and migration [1]. In bone, sema4D is expressed by the osteoclasts and binds to its receptor plexin-B1 on the osteoblasts, resulting in decreased bone formation through suppression of IGF-1 signaling and modulation of osteoblast motility [2]. Additionally, sema4D delayed osteoclast differentiation and affected osteoclast function in vitro [3]. Furthermore, a sema4D-specific antibody promoted bone formation in a co-culture system of human osteoclasts and osteoblasts; injection of this antibody, in either prophylactic or therapeutic doses in ovariectomized mice, was protective against bone loss by promoting osteoblastic bone formation without affecting osteoclastic bone resorption [2]. These data suggest that sema4D may represent a promising therapeutic target for bone diseases [2].

Current antiosteoporotic treatments include antiresorptive agents, such as zoledronic acid [4] and denosumab [5], which target the osteoclast, thereby inhibiting bone resorption and osteoanabolic agents, such as teriparatide [6], which target the osteoblast, thereby stimulating bone formation. The decrease in bone resorption induced by either zoledronic acid or denosumab is soon accompanied by a decrease in bone formation, due to the coupling effect. Accordingly, increased bone formation by teriparatide is followed by an increase in bone resorption.

To date, there are no data for circulating Sema4D/plexin-B1 levels in postmenopausal osteoporosis. Furthermore, therapies that affect osteoclast or osteoblast number and activity, such as zoledronic acid, denosumab or teriparatide, could influence sema4D expression and subsequently its serum levels.

The main aims of this prospective study were to evaluate: i) circulating sema4D and plexin-B1 levels in postmenopausal women with normal and low bone mass; and ii) the 3-month effect of antiresorptives, such as zoledronic acid and denosumab, and osteoanabolics, such as teriparatide, on sema4D and plexin-B1 levels. Secondary aim was the evaluation of potential correlations between sema4D or plexin-B1 and relevant clinical or laboratory parameters.

2. Patients and methods

2.1 Patients

This was an interventional, parallel assignment, open-label clinical trial. Patients were recruited at the outpatient clinics for Metabolic Bone Diseases of 424 General Military Hospital, Thessaloniki, Greece. Postmenopausal Caucasian women with low bone mass (bone mineral density [BMD] T-score of ≤ -2.0 at the lumbar spine [LS] and/or the non-dominant femoral neck [FN]) were recruited as patients. Patients were assigned to either zoledronic acid 5 mg i.v. (zoledronic acid group) or denosumab 60 mg s.c. (denosumab group) or teriparatide 20 μg/day s.c. for 3 months (teriparatide group). There was no randomization; patients were assigned to each treatment according to current national guidelines [7]. Women with BMD T-score of > -1.5 at the LS and the non-dominant FN were recruited as controls (control group). Patients of all groups and controls were selected to be matched for age, age at menopause and body mass index (BMI).

Exclusion criteria for all groups were: i) age < 40 years; ii) any bone and mineral disorder other than osteoporosis, including primary or secondary hyperparathyroidism, Paget’s disease of bone, osteogenesis imperfecta, rheumatologic diseases, paraplegia, chronic immobilization; iii) previous treatment with zoledronic acid, denosumab, teriparatide or strontium ralenate; iv) severe liver or kidney disease (creatinine clearance < 60 ml/min/1.73 m²) or liver or kidney transplantation; v) uncontrolled thyroid disease; vi) any malignancy; and vii) history or concomitant medications that could affect bone metabolism. All patients received 1000 mg/day of calcium carbonate and 800 IU/day of vitamin D throughout the study. Controls did not receive calcium/vitamin D supplements.

The study was approved by the local ethics committee and was in accordance with the Declaration of Helsinki and the International Conference on Harmonization for Good Clinical Practice. Written informed consent was obtained from all participants.

2.2 Methods

At baseline, a detailed history including fracture history and smoking habits was obtained, physical examination and BMD measurement at the LS and/or the non-dominant FN were performed and BMI was calculated. Morning (8 – 9 am) blood samples after an overnight fast were obtained from all the participants at baseline and 3 months after zoledronic acid infusion, denosumab injection, teriparatide initiation or no treatment for zoledronic acid, denosumab, teriparatide and control group, respectively. Serum levels of total calcium, albumin (for calcium correction), phosphate, creatinine and total alkaline phosphatase (tALP) were measured within an hour from blood drawing. Additional samples were centrifuged immediately and serum was separated and stored at -80°C for the measurement of the remaining study’s parameters, which was performed in one batch at the end of the study. These parameters were: 25-hydroxyvitamin D (25(OH) D; ECLIA, Elecsys Vitamin D3, Roche Diagnostics, Mannheim, Germany; sensitivity 7.5 nmol/l; intra-assay coefficient of variation [CV] 1.7 – 7.8%; inter-assay CV 2.2 – 10.7%), intact parathyroid hormone (PTH; ECLIA, Elecsys PTH intact, Roche Diagnostics, Mannheim, Germany; sensitivity 1.20 pg/ml; intra-assay CV 1.5 – 2.7%; inter-assay CV 3.0 – 6.5%), serum carboxyterminal telopeptide of type 1 collagen (CTx; chemiluminescence immunoassay, ImmunoDiagnostic Systems, Herlev, Denmark; intra-assay CV 1.7 – 3.0%; inter-assay CV 2.5 – 10.9%), bone-specific alkaline phosphatase (bALP; EIA, Ostase BAP, ImmunoDiagnostic Systems, Herlev, Denmark; sensitivity 1.0 μg/l; intra-assay CV 1.5 – 2.7%; inter-assay CV 3.0 – 6.5%), plexin-B1 (ELISA; Uscn Life Science, Inc., Wuhan, China; sensitivity 0.59 ng/ml; intra-assay CV < 10%; inter-assay CV < 12%) and sema4D (ELISA; Human SEMA4D/CD100, Uscn Life Science, Inc., Wuhan, China; sensitivity 0.39 ng/ml; intra-assay CV < 8%; inter-assay CV < 10%). Corrected calcium was calculated by the formula: Corrected calcium (mg/dl) = Ca (mg/dl) + 0.8 × (4.0 – albumin [g/dl]). BMD was measured at baseline by dual energy X-ray absorptiometry (DXA) using a DPX-IQ densitometer (Lunar Corp., Madison, WI, USA).

2.3 Statistical analysis

Data for continuous variables are presented as mean ± standard error of the mean (SEM). Data for categorical variables are presented as numbers and/or percentages. Kolmogorov-Smirnov test was used to test the normality of distribution of continuous variables. Paired T-test or Wilcoxon Signed Ranks test were used to test for differences within the levels of continuous variables. Independent samples T-test or Mann-Whitney test were used for between-group comparisons, in cases of two groups of continuous variables. One-way analysis of variance (ANOVA) or Kruskal-Wallis test were used in cases of more than two groups of continuous variables. In case of statistically significant difference in ANOVA or Kruskal-Wallis test, Tukey’s post-hoc adjustment was used for multiple pairwise comparisons. Chi-square or Fischer’s exact test were used for between-group differences in categorical variables. Spearman’s (rs) coefficient of correlation was used for bivariate correlations between continuous variables. Partial coefficient (r_p) was used for binary correlations adjusted for co-founders. Multiple linear regression analysis was used to search for independent associates of serum sema4D, plexin-B1 or their ratio at baseline or month 3. For the need of regression analyses, variables not following a normal distribution were logarithmically transformed. A two-sided p-value of < 0.05 was considered statistically significant in all the aforementioned tests. Statistical analysis was performed by SPSS 21.0 for Macintosh (IBM Corp., Armonk, NY).

3. Results

A total of 120 Caucasian postmenopausal women (90 patients and 30 controls) were initially recruited. Out of them, 118 patients completed the study, whereas 2 women of the teriparatide group withdrew their consent approximately 1 week and 1.5 month after baseline. Comparative anthropometric and baseline DXA data are presented in Table 1. Groups were matched for age, age at menopause and BMI. BMI remained essentially unchanged at month 3 in all study groups. As expected, the control group had higher LS and FN BMD and T-scores than all patient groups. In the sum of the 3 groups of patients, 25 (27.8%) patients had previously received bisphosphonate treatment other than zoledronic acid, with higher rates in the teriparatide group. There was no statistically significant difference in current cigarette smoking between sum of the 3 groups of patients, despite a trend towards higher rates in the teriparatide group.

Table 1. Comparative baseline data of the study groups.

Variable	Control (n = 30)	Zoledronic acid (n = 30)	Denosumab (n = 30)	Teriparatide (n = 28)	p-value for trend between groups
Age (years)	64.6 ± 1.4	66.1 ± 1.9	66.4 ± 1.4	65.4 ± 1.5	0.857
Age at menopause (years)	49.7 ± 0.7	48.3 ± 0.9	49.0 ± 0.9	47.8 ± 1.0	0.428
BMI (kg/m^2)	30.38 ± 0.95	28.05 ± 1.05	29.93 ± 0.61	28.98 ± 0.87	0.247
LS BMD (g/cm^2)	1.101 ± 0.024	0.921 ± 0.023*	0.925 ± 0.029*	0.839 ± 0.018*	< 0.001
LS T-score (SD)	-0.65 ± 0.20	-2.23 ± 0.18*	-2.17 ± 0.21*	-2.82 ± 0.12*	< 0.001
FN BMD (g/cm^2)	0.880 ± 0.016	0.704 ± 0.036*	0.751 ± 0.020*	0.660 ± 0.044*	0.002
FN T-score (SD)	-1.27 ± 0.11	-3.05 ± 0.30*	-2.62 ± 0.16*	-3.39 ± 0.34*	< 0.001
Previous BP treatment; n (%)	0 (0)	8 (27)*	3 (10)	14 (50)*^,§	< 0.001
Low-energy fractures; n (%)	0 (0)	4 (13)*	7 (23)*	13 (46)*^,‡	< 0.001
Current smoking; n (%)	3 (10)	2 (7)	6 (20)	9 (32)	0.057

Data are presented as mean ± standard error of the mean (SEM) or absolute numbers (%).

*Compared to control group.

^‡Compared to zoledronic acid group.

^§Compared to denosumab group (Tukey’s post-hoc adjustment for between-group pairwise comparisons).

BMD: Bone mineral density; BMI: Body mass index; BP: Bisphosphonate; FN: Femoral neck; LS: Lumbar spine.

Comparative baseline and 3-month biochemical data of the study groups are presented in Table 2. Serum sema4D significantly increased within the denosumab group, whereas it decreased within the teriparatide group; however, it also decreased within the control group (Table 2). Serum plexin-B1 significantly decreased only within the teriparatide group. The sema4D/plexin-B1 ratio was significantly increased only within the denosumab group. The changes in bone turnover markers and other parameters related to calcium homeostasis followed an expected pattern, apart from an unexpected increase in bALP and decrease in 25(OH)D within the control group (Table 2).

Table 2. Comparative baseline and 3-month biochemical data of the study groups.

Variable	Group	Baseline	Month 3	p-value within groups
sema4D (ng/ml)	Control	27.5 ± 3.1	19.7 ± 3.0	0.026
	Zoledronic acid	51.4 ± 7.1*	55.1 ± 6.2*	0.773
	Denosumab	11.1 ± 1.4^‡	17.0 ± 2.9^‡	0.026
	Teriparatide	58.3 ± 6.0*^,§	46.3 ± 5.6*^,§	0.013
p-value between groups		< 0.001	< 0.001
Plexin-B1 (ng/ml)	Control	4.9 ± 1.5	3.0 ± 1.1	0.223
	Zoledronic acid	2.5 ± 0.6	4.9 ± 1.6	0.102
	Denosumab	3.0 ± 1.3	1.5 ± 0.4	0.090
	Teriparatide	4.7 ± 0.4	3.3 ± 0.5	0.017
p-value between groups		0.300	0.113
sema4D/plexin-B1	Control	28.8 ± 6.2	22.4 ± 8.1	0.685
	Zoledronic acid	46.4 ± 9.9	65.5 ± 16.3	0.428
	Denosumab	16.6 ± 2.9^‡	58.2 ± 23.5	0.004
	Teriparatide	16.8 ± 3.2^‡	19.8 ± 2.8	0.362
p-value between groups		0.003	0.126
tALP (U/l)	Control	82.9 ± 3.8	82.6 ± 3.9	0.717
	Zoledronic acid	80.0 ± 4.2	62.2 ± 3.3*	< 0.001
	Denosumab	83.0 ± 3.9	54.4 ± 3.0*	< 0.001
	Teriparatide	63.4 ± 3.6*^,‡,§	74.4 ± 5.1*	0.002
p-value between groups		0.001	< 0.001
bALP (μg/l)	Control	19.4 ± 1.1	31.8 ± 3.2	0.001
	Zoledronic acid	28.4 ± 2.0*	21.4 ± 2.6*	< 0.001
	Denosumab	20.4 ± 1.2^‡	18.7 ± 2.4*	0.165
	Teriparatide	17.5 ± 1.9^‡	21.3 ± 1.4^§	0.005
p-value between groups		< 0.001	0.002
CTx (ng/ml)	Control	0.454 ± 0.026	0.440 ± 0.030	0.205
	Zoledronic acid	0.412 ± 0.042	0.128 ± 0.016*	< 0.001
	Denosumab	0.497 ± 0.028	0.100 ± 0.012*	< 0.001
	Teriparatide	0.433 ± 0.040	0.929 ± 0.130*^,‡,§	< 0.001
p-value between groups		0.346	< 0.001
PTH (pg/ml)	Control	35.1 ± 3.3	29.7 ± 1.8	0.073
	Zoledronic acid	24.9 ± 2.4*	30.9 ± 3.3	0.024
	Denosumab	32.3 ± 2.6	36.1 ± 4.6	0.558
	Teriparatide	19.8 ± 2.2*^,§	13.7 ± 1.9*^,‡,§	0.003
p-value between groups		< 0.001	< 0.001
25(OH)D (ng/ml)	Control	39.2 ± 2.2	28.6 ± 2.4	< 0.001
	Zoledronic acid	28.8 ± 2.2*	27.2 ± 1.6	0.496
	Denosumab	37.4 ± 1.9^‡	34.8 ± 2.0^‡	0.203
	Teriparatide	21.5 ± 1.9*^,§	32.7 ± 2.2	< 0.001
p-value between groups		< 0.001	0.030
Corrected calcium (mg/dl)	Control	9.2 ± 0.1	9.2 ± 0.1	0.721
	Zoledronic acid	9.2 ± 0.1	9.0 ± 0.1	0.001
	Denosumab	9.2 ± 0.1	9.0 ± 0.1	0.082
	Teriparatide	9.0 ± 0.1	9.2 ± 0.1	0.021
p-value for between groups		0.441	0.115
Phosphate (mg/dl)	Control	3.5 ± 0.1	3.5 ± 0.1	0.436
	Zoledronic acid	3.4 ± 0.1	3.3 ± 0.1	0.244
	Denosumab	3.5 ± 0.1	3.4 ± 0.1	0.141
	Teriparatide	3.6 ± 0.1	3.7 ± 0.1^‡	0.446
p-value for between groups		0.053	0.017
Creatinine (mg/dl)	Control	0.86 ± 0.02	0.87 ± 0.03	0.825
	Zoledronic acid	0.90 ± 0.02	0.90 ± 0.02	0.450
	Denosumab	0.84 ± 0.03	0.80 ± 0.03	0.277
	Teriparatide	0.82 ± 0.05	0.76 ± 0.04 ^‡	0.022
p-value for between groups		0.275	0.013

Data are presented as mean ± standard error of the mean (SEM).

*Compared to control group.

^‡Compared to zoledronic acid group.

^§Compared to denosumab group (Bonferroni post-hoc adjustment for between groups pairwise comparisons).

25(OH)D: 25-hydroxyvitamin D; bALP: Bone-specific alkaline phosphatase; CTx: Serum carboxyterminal telopeptide of type 1 collagen; PTH: Intact parathyroid hormone; Sema4D: Semaphorin-4D; tALP: Total alkaline phosphatase.

Between patients (n = 88) at baseline, circulating sema4D and plexin-B1 were higher in those previously being on bisphosphonate treatment as compared to naïve patients (62.9 ± 8.4 vs 30.7 ± 3.6 ng/ml; p < 0.001 and 4.1 ± 0.5 vs 3.1 ± 0.7 ng/ml, respectively; p = 0.001). There was no difference in sema4D/plexin-B1 ratio between groups (23.4 ± 5.5 vs 28.2 ± 5.1; p = 0.694). Similar data were retrieved when the control group was entered into this comparison. There were no between-group differences in sema4D, plexin-B1 or their ratio between those with previous low-energy fracture or not, or between smokers and non-smokers (both in patients and in the sum of the participants).

Bivariate correlations of sema4D, plexin-B1 and their ratio with other study parameters in the sum of participants at baseline are presented in Table 3. Notably, sema4D and plexin-B1 were positively correlated with each other. Furthermore, sema4D was inversely correlated with CTx, PTH and 25(OH)D, whereas there was a trend towards inverse correlation with LS BMD (p = 0.053) and positive correlation with creatinine (p = 0.059). Plexin-B1 was inversely correlated only with 25(OH)D, whereas sema4D/plexin-B1 ratio was inversely correlated with CTx and PTH (Table 3). In multiple linear regression analysis at baseline, previous bisphosphonate treatment, log(PTH) and 25(OH)D, but not group, previous low-energy fracture, current smoking, age at menopause, BMI, LS BMD, log(plexin-B1), CTx and creatinine, were independently associated with log(sema4D) (Table 4). According to this model, previous bisphosphonate treatment, and lower PTH and 25(OH)D were independently associated with higher sema4D levels at baseline. In a similar model of regression analysis, log(plexin-B1) at baseline was independently associated with previous bisphosphonate treatment, 25(OH)D and BMI, but not group, previous low-energy fracture, current smoking, age at menopause, LS BMD, log(sema4D), CTx, log(PTH) and creatinine (Table 5). According to this model, previous bisphosphonate treatment, and lower 25(OH)D and BMI were independently associated with higher plexin-B1 levels at baseline. Sema4D/plexin-B1 ratio was not independently associated with any of the study’s parameters.

Table 3. Bivariate correlations at baseline between sema4D, plexin-B1 and their ratio with other study parameters in the sum of participants.

Variable	Sema4D (ng/ml)	Plexin-B1 (ng/ml)	Sema4D/plexin-B1
Age (years)	-0.031 (0.742)	-0.075 (0.421)	0.045 (0.628)
Age at menopause (years)	-0.175 (0.068)	-0.087 (0.363)	-0.001 (0.994)
BMI (kg/m^2)	0.004 (0.996)	-0.126 (0.176)	0.161 (0.083)
LS BMD (g/cm^2)	-0.191 (0.053)	-0.163 (0.099)	0.078 (0.434)
FN BMD (g/cm^2)	-0.229 (0.179)	-0.150 (0.383)	-0.027 (0.876)
tALP (U/l)	-0.137 (0.145)	-0.063 (0.506)	-0.045 (0.635)
bALP (μg/l)	-0.002 (0.981)	-0.060 (0.517)	0.032 (0.736)
CTx (ng/ml)	-0.282 (0.002)	0.040 (0.670)	-0.261 (0.005)
PTH (pg/ml)	-0.388 (< 0.001)	-0.066 (0.484)	-0.204 (0.029)
25(OH)D (ng/ml)	-0.316 (< 0.001)	-0.203 (0.030)	-0.012 (0.899)
Corrected calcium (mg/dl)	-0.060 (0.523)	-0.054 (0.564)	-0.012 (0.900)
Phosphate (mg/dl)	0.104 (0.264)	0.072 (0.441)	0.039 (0.675)
Creatinine (mg/dl)	0.193 (0.059)	-0.003 (0.978)	0.126 (0.221)
Plexin-B1 (ng/ml)	0.307 (0.001)	-	-
sema4D/plexin-B1	0.361 (0.001)	-0.767 (< 0.001)	-

Data are presented as Spearman’s coefficient of correlation; r_s (p-value).

25(OH)D: 25-hydroxyvitamin D; bALP: Bone-specific alkaline phosphatase; BMD: Bone mineral density; CTx: Serum carboxyterminal telopeptide of type 1 collagen; FN: Femoral neck; LS: Lumbar spine; PTH: Intact parathyroid hormone; Sema4D: Semaphorin-4D; tALP: Total alkaline phosphatase.

Table 4. Multiple linear regression analysis for independent associates of log(sema4D) (stepwise method)*.

Covariate	β coefficient	p-value	95% CI for B coefficient
Previous BP treatment^‡	0.243	0.016	0.047 to 0.438
Log(PTH) (pg/ml)	-0.626	0.001	-1.005 to -0.248
25(OH)D (ng/ml)	-0.007	0.033	-0.013 to -0.001

*Variables excluded from the analysis (non-independent associates) were group, previous low-energy fracture, current smoking, age at menopause, BMI, LS BMD, log(plexin-B1), CTx and creatinine.

^‡Naïve participants were rated as 0, whereas those previously on bisphosphonates were rated as 1.

25(OH)D: 25-hydroxyvitamin D; BMD: Bone mineral density; BMI: Body mass index; BP: Bisphosphonate; CTx: Serum carboxyterminal telopeptide of type 1 collagen; LS: Lumbar spinal; PTH: Intact parathyroid hormone.

Table 5. Multiple linear regression analysis for independent associates of log(plexin-B1) (stepwise method)*.

Covariate	β coefficient	p-value	95% CI for B coefficient
Previous BP treatment^‡	0.356	0.015	0.070 to 0.642
25(OH)D (ng/ml)	-0.012	0.018	-0.021 to -0.002
BMI (kg/m^2)	-0.029	0.032	-0.055 to -0.003

*Variables excluded from the analysis (non-independent associates) were group, previous low-energy fracture, current smoking, age at menopause, LS BMD, log(plexin-B1), PTH, CTx and creatinine.

^‡Naïve participants were rated as 0, whereas those previously on bisphosphonates were rated as 1.

25(OH)D: 25-hydroxyvitamin D; BMD: Bone mineral density; BMI: Body mass index; BP: Bisphosphonate; CTx: Serum carboxyterminal telopeptide of type 1 collagen; LS: Lumbar spinal; PTH: Intact parathyroid hormone.

Partial correlations (adjusted for group) between sema4D, plexin-B1 and their ratio, and other study parameters in the sum of participants at month 3 are presented in Table 6. Sema4D was not correlated with plexin-B1. Sema4D was positively correlated with creatinine, inversely correlated with age at menopause and PTH, whereas there was a marginally non-significant, inverse correlation between sema4D and 25(OH)D, (p = 0.051). Plexin-B1 was not correlated with any other parameter at month 3, whereas sema4D/plexin-B1 ratio was correlated with bALP. In multiple linear regression analysis at month 3, log(sema4D) was independently associated with previous bisphosphonate treatment (β = 24.4; 95% CI = 9.8 – 39.0; p = 0.001) and 25(OH)D (β = -0.794; 95% CI = -1.302 to -0.285; p = 0.003), but not current treatment, previous low-energy fracture, current smoking, age at menopause, BMI, log(plexin-B1), CTx, log(PTH) and creatinine. Neither plexin-B1 nor sema4D/plexin-B1 ratio were independently associated with any of the study’s parameters in regression analysis at month 3.

Table 6. Partial correlations at month 3 between sema4D, plexin-B1 and their ratio with other study parameters, after adjustment for group in the sum of participants.

Variable	Sema4D (ng/ml)	Plexin-B1 (ng/ml)	Sema4D/plexin-B1
Age (years)	-0.016 (0.868)	0.127 (0.187)	0.078 (0.420)
Age at menopause (years)	-0.271 (0.005)	0.022 (0.828)	-0.055 (0.581)
BMI (kg/m^2)	-0.024 (0.804)	0.163 (0.091)	0.110 (0.254)
tALP (U/l)	0.087 (0.373)	0.145 (0.138)	-0.016 (0.872)
bALP (μg/l)	-0.025 (0.796)	0.026 (0.790)	0.263 (0.006)
CTx (ng/ml)	0.023 (0.814)	0.023 (0.813)	-0.143 (0.135)
PTH (pg/ml)	-0.281 (0.003)	0.022 (0.822)	-0.089 (0.363)
25(OH)D (ng/ml)	-0.190 (0.051)	-0.133 (0.175)	-0.072 (0.465)
Corrected calcium (mg/dl)	0.014 (0.885)	-0.049 (0.613)	0.022 (0.823)
Phosphate (mg/dl)	0.185 (0.057)	-0.065 (0.504)	0.034 (0.729)
Creatinine (mg/dl)	0.207 (0.043)	0.092 (0.374)	-0.148 (0.151)
Plexin-B1 (ng/ml)	0.121 (0.207)	-	-
Sema4D/plexin-B1	0.161 (0.094)	-0.223 (0.019)	-

Data are presented as coefficient of partial correlation; r_p (p-value).

25(OH)D: 25-hydroxyvitamin D; bALP: Bone-specific alkaline phosphatase; BMD: Bone mineral density; BMI: Body mass index; CTx: Serum carboxyterminal telopeptide of type 1 collagen; PTH: Intact parathyroid hormone; Sema4D: Semaphorin-4D; tALP: Total alkaline phosphatase.

4. Discussion

Sema4D, expressed by the osteoclasts, may mediate the crosstalk between the osteoblast and the osteoclast during bone remodeling, through binding to its receptor plexin-B1 on the osteoblasts, resulting in reduced bone formation [8]. In this study in humans we showed, for the first time in the literature, that antiresorptive agents tend to increase while osteoanabolic agents tend to decrease circulating sema4D in postmenopausal women with low bone mass; this mechanism may mediate the coupling between formation and resorption that limits the net increase in bone mass with all currently available antiosteoporotic agents.

In our attempt to evaluate the effect of antiosteoporotic treatment on circulating sema4D, we used the most potent commercially available antiresorptive and osteoanabolic agents, namely, zoledronic acid, denosumab and teriparatide. Bone turnover markers, CTx, bALP and tALP, were altered as expected following each treatment, providing evidence that all agents used had the anticipated effect on bone turnover.

Circulating sema4D levels were not significantly affected by zoledronic acid, but were increased by denosumab. Given that sema4D is abundantly expressed in the immune cells [1], this difference could ensue from the unique effects of denosumab on lymphocyte-derived receptor activator of nuclear factor κB ligand (RANKL) and the RANKL-mediated effects on the immune response. Another explanation for this discrepancy could be that the two antiresorptives do not share the same effect on bone-derived sema4D, given that they have a distinct mechanism of inflicting the osteoclast [9]. However, previous bisphosphonate treatment was also independently associated with higher sema4D levels in our study; since fewer patients had previously received bisphosphonates in the denosumab group compared to either the zoledronic acid or the teriparatide group, this may partly explain: i) the lower baseline sema4D levels in the denosumab compared to the other two groups; and ii) the distinct effect of denosumab and zoledronic acid on sema4D levels at 3 months. Furthermore, a distinct short-term and long-term effect of bisphosphonates on sema4D levels may be speculated, since zoledronic acid did not significantly alter sema4D levels at 3 months while long-term use of bisphosphonates seems to have a more prominent effect. On the other hand, sema4D levels were decreased in the teriparatide group. Although sema4D levels were also decreased in the control group, there seems to exist an opposite effect of antiresorptives, mainly denosumab, and osteoanabolics, namely, teriparatide, on circulating sema4D.

It is of note that, the change in circulating levels of sema4D following each treatment was opposite than the expected according to the ensuing osteoclast activity and number. In specific, sema4D levels increased within patients with the greater suppression of osteoclast activity (denosumab group) while decreased in case of osteoclasts’ induction by teriparatide. This pattern was further supported by the higher sema4D levels in those patients previously treated with bisphosphonates, and thus with an already suppressed osteoclastic activity compared to naïve patients. We cannot provide a solid explanation for this finding. We could only speculate that circulating sema4D could also be derived from another cell population besides the osteoclasts, such as T cells, which are known producers of sema4D [1], or osteocytes, which are already known to control osteoblasts through sclerostin production. Alternatively, ‘dying’, due to the antiresorptive treatment, osteoclasts could release a burst of sema4D. Regardless of the source, this putative sema4D production could offer an additional suppression of the osteoblasts, if needed, during antiresorptive treatment, in an attempt to maintain coupling; on the other hand, during osteoanabolic treatment osteoclasts and other sema4D producing cells might cease to produce sema4D, thus allowing osteoblast activity and bone formation. In accordance with the latter speculation, a sema4D-targeting molecule increased osteoblast number and bone volume in an animal model of osteoporosis [10].

The increase of circulating sema4D in pre-treated with bisphosphonates patients as well as following denosumab might imply a role in the oversuppression of bone turnover that leads to osteonecrosis of the jaw (ONJ) [11] and atypical fractures [12] in bisphosphonate- and denosumab-treated patients. However, the hypothesis that sema4D might represent a risk factor or a predictor of ONJ and atypical fractures in antiresorptive-treated patients is highly speculative and needs to be tested in larger studies.

In our population at baseline, a trend towards an inverse correlation with BMD was found. This implies that women with lower BMD may have higher sema4D levels. The existence of such a correlation should be verified or excluded in larger studies.

Women in the teriparatide group had lower 25(OH)D levels at baseline compared to the other groups. Although we have no solid explanation for this, we could speculate that these women with lower BMD and more low-energy fractures were avoiding outdoor activities and, therefore, sun exposure. Furthermore, in the teriparatide group 25(OH)D levels increased significantly after 3 months of treatment, and this may exert an independent effect on sema4D levels as regression analysis indicated.

Our study has certain limitations. First, it was an open-label study; however, given the different routes of administration for zoledronic acid, denosumab and teriparatide, we could not easily achieve blindness. Second, no randomization was performed at baseline; patients were allocated to each treatment according to current guidelines, thereby committing a selection bias. Third, the study targeted the short-term effect of treatment; longer evaluation could verify the changes observed at 3 months and provide information regarding long-term behavior of sema4D/plexin-B1 post treatment. Fourth, the fact that previous bisphosphonate treatment was not in the exclusion criteria was also a drawback; however, judging it a posteriori, it provided us indirect information regarding long-term bisphosphonate use. Fifth, 25(OH)D decreased at 3 months in our control group and bALP increased without a respective increase in CTx or even tALP. We do not have a clear explanation for these changes. The lack of supplementation with calcium/vitamin D preparations in the controls could partially explain the decrease of 25(OH)D. Finally, it is not known to what extent serum levels of sema4D/plexin-B1 reflect their expression in bone or whether release from other tissues contributes to sema4D/plexin-B1 circulating levels; however, we administered agents that mainly affect the skeleton, therefore, it is likely that the observed changes in sema4D/plexin-B1 levels are attributed to altered release from the bone.

In conclusion, our study represents the first exploratory attempt to evaluate circulating sema4D and plexin-B1 levels in osteoporosis in humans. Our data suggest that antiresorptive and osteoanabolic agents may exert opposite effects on circulating sema4D levels in postmenopausal women with low bone mass and thus, sema4D may mediate the coupling effect that occurs following treatment. This study, in addition to existing in vitro data [2], warrants further studies to evaluate the role of sema4D in osteoporosis, shed light on the mechanism of action of antiosteoporotic agents and trigger research for future anti-osteoporotic drugs [10].

Acknowledgment

We thank Roche Diagnostics Hellas for providing us the kits for the measurements of PTH and 25(OH)D.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.