Associations of serum sclerostin with bone mineral density, markers of bone metabolism and thalassaemia characteristics in adult patients with transfusion-dependent beta-thalassaemia

Abstract

Aim of the study: To assess serum sclerostin in transfusion-dependent beta-thalassaemia patients versus healthy controls and to examine its associations with bone mineral density, bone metabolism markers and beta thalassaemia alterations.

Material and methods: Sixty-two transfusion-dependent beta-thalassaemia (TDßT) patients and 30 healthy controls were evaluated for serum sclerostin, osteocalcin, beta-cross laps, osteoprotegerin and serum level of receptor activator of nuclear factor kappa-Β ligand (sRANKL). Bone mineral density was measured at the lumbar spine and femoral neck. Thalassaemia characteristics were collected from the patients’ medical records.

Results: A significantly higher sclerostin level (median 565.50 pmol/L) was observed in the transfusion-dependent beta-thalassaemia patients vs. the healthy controls (median 48.65 pmol/L, p < .001). Sclerostin showed significant associations with the Z-scores at the lumbar spine and femoral neck, osteocalcin, beta-cross laps, osteoprotegerin, sRANKL, pretransfusion haemoglobin, liver iron concentration and female gonadal state. Significantly higher levels of sclerostin were observed in splenectomized TDßT patients and in those with fragility fractures. Age, sex, body mass index, disease severity, serum ferritin, cardiac T2* and male gonadal state did not show significant associations with sclerostin.

Conclusion: Sclerostin may play a role in the bone pathophysiology of beta-thalassaemia patients and could serve as a marker of severe osteoporosis.

KEY MЕSSAGES

• Serum sclerostin is more than 10-fold higher in adult patients with transfusion-dependent beta-thalassaemia compared to healthy controls.

• Serum sclerostin is negatively associated with bone mineral density and the bone synthesis markers and positively with the bone resorption indices.

• Serum sclerostin is significantly associated with pre-transfusion haemoglobin, liver iron concentration, splenectomy status and fragility fracture events in adult patients with transfusion-dependent beta-thalassaemia.

• Serum sclerostin could serve as a marker of severe osteoporosis in beta-thalassaemia patients.

Introduction

Beta-thalassaemia is a hereditary hemopathy, resulting from a genetically-determined quantitative defect in the synthesis of beta-globin chains. The leading clinical characteristics of the disease include anaemia of different severity, ineffective erythropoiesis with extramedullary involvement, and secondary iron overload due to regular transfusions. Hypertansfusion treatment and chelating therapy have helped alleviate these manifestations to a great extent, consequently leading to patients’ improved quality of life and increased survival rate.

However, in the long-term other morbidities are likely to appear, one of which is thalassaemia-related bone disease. Despite the recent advances in bone-bilogy, thalassaemia-related bone disease has not been sufficiently studied and understood as a distinctive condition with its own unique characteristics, including bone deformities, osteoporosis and fragility fractures [1].

One of the pivotal aims of current research about thalassaemia-related bone disease is to detect abnormalities in signalling systems, affecting bone metabolism in patients with thalassaemia and their regulatory mechanisms.The canonical Wingless and int-1 Wnt/β-catenin signalling pathway has been identified as an important regulator of osteogenesis [2–5]. This pathway controls the processes of proliferation, differentiation and apoptosis of the osteoblastic-osteocytic line [6]. It is also proposed that this pathway is significantly involved in the pathogenesis of osteoporosis in thalassaemia major patients. Sclerostin and Dickkopf-1 are considered negative modulators of this signalling system. Research on thalassemic cohorts has provided evidence about their influence on bone mineral density (BMD) in thalassaemia major patients [7–9]. Since the present study examined the role ofsclerostin in TDßT patients, the background information in the remaining part of this section does not include Dickkopf-1 and focuses on sclerostin alone.

Sclerostin is a molecule which modulates the Wnt-signalling pathway [10–11] by impeding bone formation in osteoblasts. It interplays with Wnt-signalling in different ways [11–16], including blockage of the lipoprotein-receptor (LPR) 5 and 6 on the osteocytic surface and binding with LPR-4 [17] and members of the bone morphogenic protein (BMP) family or other molecules [18]. Under normal circumstanses sclerostin is considered to be a marker of the activity of mature osteocytes, which were innitially thought be the main source of its production [19]. It was also found that sclerostin can be secreted by osteoclast precursors in bone marrow and decreases in the course of their maturation [14]. Additionally, in vitro studies have demonstrated that sclerostin can also be synthesised by mature osteoclasts [20]. These findings highlight the complexity of sclerostin synthesis and cast light on its biological activity.

Following the publications of Li et al. [21] and Lewiecki [22] who found an increased bone formation after the administration of anti-sclerostin antibodies in animal studies and early phase clinical settings there has been a growing interest in this protein in clinical settings [23–26]. Serum sclerostin has been studied in different disease states, including endocrine [27–29] inflammatory [30–32] and malignant diseases [33–35] and in relation to demographic variables, such as age and gender [36]. However, research about its role in transfusion-dependent beta thalassaemia (TDßT) patients is still rather insufficient.

In an effort to explore the potential utility of treating bone damage and other manifestations of TDßT through control of sclerostin levels, we set out to assess serum sclerostin in TDßT patients and healthy controls and to examine its association with thalassaemia-specific and bone disease parameters.

Recognising the need for further research about serum sclerostin in TDßT patients, we aimed the present study to fulfil the following goals: (1) To examine the serum levels of sclerostin in adult TDßT patients versus that of healthy controls; and (2) To seek possible associations between sclerostin levels and patients’ age, BMI, sex, BMD status, bone synthesis and resorption biochemical markers, basic beta thalassaemia alterations, gonadal and parathyroid status.

Materials and methods

Participants

A total of 76 TDßT patients monitored at the “Sveti Georgi” University Hospital, Plovdiv, Bulgaria and at the Haematology Clinic of “Prof. Dr. St. Kirkovich” University Hospital, Stara Zagora, Bulgaria were initially screened for eligibility into the study. Subsequently, fourteen patients were excluded due to one or more of the following reasons: five patients had been administered Denosumab during the past one year; six patients were with diabetes mellitus, three of whom also had hypothyroidism; two patients had chronic kidney disease; one patient declined participation in the study. Consequently, the final sample included 62 TDßT patients [47 with thalassaemia major (TM); 15 with thalassaemia intermedia (TI)] and 30 clinically healthy controls.

The healthy controls were recruited from individuals visiting the two medical institutions mentioned above for prophylactic exams. All procedures were performed in adherence to the World Medical Association Declaration of Helsinki (1964) and its revised version (2000), Edinburgh. Ethical approval was granted by the Local Ethics Committee at the Medical University of Plovdiv (31 March 2017). Informed consent in written form was obtained from each participant in the study, including patients and healthy controls, following the requirements of the WMA Declaration of Helsinki.

Inclusion and exclusion criteria

The patients were selected in view of the following inclusion criteria: (1) to be of age over 18 years; (2) to have at least electrophoretically confirmed beta-thalassaemia; (3) to be transfusion-dependendent; (4) to be naïve to osteoporosis treatment. Patients were excluded from the study in the presence of one or more of the following conditions: (1) history of familial bone disease; (2) autoimmune diseases; (3) use of corticosteroids for any reason; (4) diabetes mellitus; (5) thyroid pathology; (6) chronic inflammation; (7) chronic kidney disease

The healthy controls had to meet the following criteria for inclusion into the study: (1) to be of age over 18 years; (2) to match the study group in age and anthropometric data. The exclusion criteria were: (1) disease history of any kind; (2) family or personal history of anaemia; (3) to be active athletes; (4) history of familial bone disease; (5) smoking; (6) major trauma in the past 6 months.

Data

Demographic and anthropometric data were obtained from both TDßT patients and healthy controls, including date of birth, sex, height and weight.

Bone-methabolism markers and sex-steroids

Fasting blood samples were taken from all participants within the time period between 8 am and 9 am. The collection day for eugonadic TDßT women and those, taking hormone – replasement therapy (HRT) was designated between the 3rd and the 5th day of the menstrual cycle, in order to collect the blood samples for all the serum parameters at one time point. For hypogonadic women, non-taking HRT, and for TDßT men, there was no specific collection time requirement. The serum samples for sclerostin and bone methabolizm markers were prepared via cool-centrifugation and frozen at (−) 80 °C up to 60 min after the collection and stored until the time of the analysis. An ELISA immunoassay method was used for the measurement of the serum sclerostin levels (reactive: Human Sclerostin, ELISA kits, Mybiosource, San Diego, USA, intra and inter assay CV <8%); osteocalcin (ОСN) (reactive:Osteocalcin Instant ELISA, Affymetrix eBioscience, Austria, intra and inter assay CV <8%); Beta-cross laps (B-Ctx) (reactive: Human b-CTx, ELISA kits, Mybiosource, San Diego, USA, intra and inter assay CV <8%); osteoprotegerin (OPG) (reactive: Human Osteoprotegerin instant ELISA, eBioscience, Affymetrix company, Austria, intra-assay CV 7%; inter-assay CV 8%); sRANKL (reactive: Human ELISA kits, Biosensor, Biov.Laboratory Medicine Inc., Czech Republic, intra-assay CV 8%; inter-assay CV 11%). Oestrogen and total testosterone levels were also measured by ELISA methods.

Bone mineral density

BMD was measured by a dual-energy X-ray absorptiometry (DXA) at the lumbar spine (LS) and femoral neck (FN) with a HOLOGIC Discovery C (S/N 47070) densitometer, in compliance with the requirements of the manufacturer (CV for BMD measurement at both measured levels was 0.8%). Areal BMD in g/cm² and BMD Z-scores were calculated by the machine’s software based on the American/European Caucasian database.

Hematological parameters

The data about the patients’ pre-transfusion haemoglobin levels (g/L), peripheral erythroblasts counts (x10⁹/L), and peripheral platelet counts (x10⁹/L) were obtained from the patients’ medical records. All parameters were measured on an automatic analyser STKS Coulter, USA. The mean values of all the above parameters were calculated on a three-year retrospective basis.

Estimation of body iron burden

The serum ferritin (SF, ng/mL) was quantitatively measured by an ELISA immunoassay method (reactive: Ferritin Human of BioVendor, Czech Republic, intra-assay CV 7.3%; inter-assay CV 4.5%). The mean values were calculated on a three-year retrospective basis. Recent data (up to one-year back) about body iron burden were collected from the patients’ medical records. Liver iron concentration (LIC) was estimated on a 1.5 Tesla Magnetic Resonance (MRI). Wood’s protocol of the imaging procedure and calculation of LIC in mg/g dry weight was applied [37]. Cardiac iron overload (CIO) was also measured on a 1.5 Tesla Magnetic Resonance by the acquisition of a mid-ventricular short axis slice in a homogeneous region of interest (black-blood technique). The results were expressed in T2* in ms.

Collection of the data about fragility fractures

The data about fragility fractures was collected from the patient’s medical charts. All recorded events had happened after minor traumas. Sixteen out of nineteen patients had one fragility fracture. The remaining three patients had a history of more than one fracture event. Except one fragility fracture which was localised in the ribs, all other events were in the long bones (extremites). The rib fracture was recorded in a patient with multiple fragility fractures.

Statistical analysis

The following statistical software programmes were used to analyse and graphically illustrate the data: the Statistical Package for the Social Sciences (SPSS), Version 25 (2017); Minitab Version 18.1 (2017); and MedCalc Statistical Software version 18.11.3 (2019). We examined the data for normality through the Shapiro-Wilk’s test and took into consideration the values of skewness. Continuously measured and normally distributed variables were described through the mean values and standard deviations (±SD). Non-normally distributed variables, with skewness exceeding the limits of −1/+1, were presented as medians and the minimum and maximum values (min.-max.). Binary and ordinal data were processed in frequencies and percentages. An independent-samples t-test was used for two-group comparisons on normally distributed continuous variables and the Mann–Whitney U test for non-normally distributed variables. Associations between binary and ordinal variables were examined through a χ² test, whereas Fisher’s exact test was used for comparison of proportions in small-size subgroups. Spearman’s rank-order correlation was employed to examine the relationship between non-normally distributed and/or ordinal variables. ROC curve analysis was used to assess the diagnostic ability of serum sclerostin as an indicator of past fracture events. All p-values were two-tailed and interpreted according to the given ranges: p ≤ .05 – ≤.01 weak evidence; p < .01 – >.001 strong evidence; p < .001 very strong evidence.

Results

Demographic and clinical data

In the preliminary screening of the data, we performed multiple comparative analyses between the male and female TDßT patients and the male and female healthy controls on pertinent background parameters in order to rule out a potential confounding influence of the participants’ sex on the subsequent results and conclusions of the study. Table 1 provides the mean/median values, p-values, and 95% CI-s of the differences in means or medians. All comparsions showed a lack of significant between-sex differences (p > .05). On this ground, the male and female TDßT patients and the male and female healthy controls were joined together to form two study groups, TDßT patients vs. healthy controls.

Table 1.

Between-sex comparisons on background parameters.

	TDßT patients		Healthy controls
Parameter	Male (n = 32)	Female (n = 30)	Male (n = 15)	Female (n = 15)
Age
Mean (SD)	32.03 (12.99)	30.73 (10.60)	33.53 (11.03)	32.87 (10.73)
P (95% CI diff)	0.669 (−7.34 to 4.75 )		0.868 (−8.8 to 7.4)
Beta-CTx pg/mL
Median (IQR)	8360 (7175)	8110 (10191)	238 (261)	333 (431)
p (95% CI diff)	0.972 (−3082 to 3000 )		0.158 (−244 to 44 )
OCN ng/mL
Median (IQR)	4.90 (3.55)	4.20 (3.85)	12 (2.20)	12.40 (2.00)
p (95% CI diff)	0.657 ( −0.80 to 1.4 )		0.834 (−1.0 to 1.0)
OPG pmol/L
Median (IQR)	5.34 (5.68)	4.15 (6.88)	2.84 (3.46)	2.18 (1.83)
p (95% CI diff)	0.944 (−1.9 to 1.8 )		0.372 (−1.7 to 0.81 )
sRANKL pmol/L
Median (IQR)	5.30 (4.60)	6.40 (4.85)	3.40 (4.40)	4.10 (1.80)
p (95% CI diff)	0.461 (−2.4 to 0.70)		0.708 (−2.3 to 2.7)
Z-score LS (L1–L4)
Мean (SD)	−2.70 (1.12)	−2.56 (1.07)	−0.14 (0.72)	−0.10 (0.95)
p (95% CI diff)	0.610 (−0.41 to 0.70 )		0.912 (−0.69 to 0.77)
BMD LS (L1–L4) g/cm2
Мean (SD)	0.792 (0.12)	0.793 (0.13)	1.11 (0.08)	1.08 (0.14)
p (95% CI diff)	0.992 (−0.065 to 0.064)		0.419 ( −0.12 to 0.05)
Z-score at FN
Mean (SD)	−1.35 (1.11)	−1.53 (0.74)	0.40 (0.49)	0.15 (0.39)
p (95% CI diff)	0.461 (−0.66 to 0.30)		0.522 (−1.07 to 0.57)
BMD at FN g/cm2
Mean (SD)	0.73 (0.11)	0.69 (0.13)	0.95 (0.07)	0.87 (0.20)
p (95% CI diff)	0.196 (−.010 to 0.02)		0.195 (−.019 to 0.04)
Hb g/L
Median (IQR)	76.50 (19.00)	78.00 (15.00)
p (95% CI diff)	0.309 (−7.0 to 3.0 )
ERBL × 109/L
Median (IQR)	22.50 (28.00)	18.00 (23.00)
p (95% CI diff)	0.297 (−11 to 4.3)
SF ng/mL
Median (IQR)	950 (1690)	1200 (2400)
p (95% CI diff)	0.933 (−620 to 420 )
LIC mg/g d. w.
Median (IQR)	12.95 (14.60)	14.70 (11.10)
p (95% CI diff)	0.143 (−0.77 to 11)
Cardiac T2* ms
Median (IQR)	30.40 (16.30)	32.00 (12.50)
p (95% CI diff)	0.767 (−3.5 to 3.2 )

IQR: interquartile range; Beta-CTx: Beta C-terminal telopeptide; OCN-osteocalcin; OPG: osteoprotegerin; sRANKL: Receptor activator of NF-kB ligand; BMD – bone mineral density; Hb-haemoglobin; ERBL: Erythroblasts; SF: serum ferritin; LIC: liver iron concentration; Cardiac T2*-cardiovascular magnetic resonance T2*.

Demographic and clinical data about the 62 TDßT patients and 30 healthy controls are summarised in Table 2. The two groups were similar in age, with a mean age difference of 1.79 years (95% CI: −3.27 to 6.86). The TDßT patients’ mean age was 31.40 ± 11.81, with an age-range between 18 and 60 years. The healthy controls’ mean age was 33.20 ± 10.69, ranging between 19 and 56 years. The sex distribution was also similar, with 51.60% (32/62) males and 48.4% females (30/62) in the patient group, and 50% (15/30) males and 50% females (15/30) in the healthy controls, p = 0.885. The patients and healthy controls had comparable BMI.

Table 2.

Demographic and clinical data about the TDT patients and healthy controls.

Parameter	Main groups
	TDßT patients (n = 62)	Healthy Controls (n = 30)	95% CI difference mean/median and/or p-value
Age
Mean (SD)	31.40 (11.80)	33.20 (10.70)	1.79 (−3.27 to 6.86)
Sex n (%)
Male	32 (51.6)	15 (50)
Female	30 (48.4)	15 (50)	p = .885
BMI kg/m2
Mean (SD)	21.96 (2.60)	22.00 (2.80)	0.04 (−1.14 to 1.22)
Beta-CTx pg/mL
Median (IQR)	8120 (8566)	262 (317)	p < .001; 7858 (6575–9982)
OCN ng/mL
Median (IQR)	4.40 (3.60)	12.40 (2.00)	p < .001; −8 (−8.20 to −6.20)
OPG pmol/L
Median (IQR)	4.30 (6.32)	2.40 (2.00)	p < .001; 1.9 (1.30–3.70)
sRANKL pmol/L
Median (IQR)	5.70 (4.80)	4.10 (3.50)	p < .001; 1.6 (0.90–3.60)
Z-score LS (L1–L4)
Мean (SD)	−2.60 (1.10)	−0.10 (0.90)	p < .001; −2.50 (−2.98 to −2.03)
BMD LS (L1–L4) g/cm2
Мean (SD)	0.79 (0.12)	1.10 (0.12)	p < .001; −0.31 (−.036 to −0.25)
Z-score at FN
Mean (SD)	−1.44 (0.94)	0.28 (0.95)	p < .001;−1.72 (−0.30 to −0.14)
BMD at FN g/cm2
Mean (SD)	0.71 (0.12)	0.91 (0.15)	p < .001; −0.2 (−0.25 to −0.14)

IQR: interquartile range; BMI-body mass index; Beta-CTx: Beta C-terminal telopeptide; OCN-osteocalcin; OPG: osteoprotegerin; sRANKL: Receptor activator of NF-kB ligand; BMD – bone mineral density; p ≤ .05 – ≤.01 weak evidence; p < .01 – >.001 strong evidence; p < .001 very strong evidence.

The groups differed significantly on all other parameters included in Table 1. The TDßT patients had significantly higher serum levels of B-Ctx (p < .001), OPG (p < .001), and sRANKL (p < .001). On the other hand, they showed significantly lower levels of OCN (p < .001), Z-score at the lumbar spine (p < .001), BMD at the lumbar spine (p < .001), Z-score at the femoral neck (p < .001), and BMD at the femoral neck (p < .001).

Clinical data about the TDßT patients and patient subgroups

The TDßT sample included 47 patients with TM, of whom 24 (51%) were female and 23 (49%) male; and 15 patients with TI, among whom 6 (40%) female and 9 (60%) male. Clinical data about the whole group of TDßT patients and the two patient subgroups are presented in Table 3. The patient subgroups were compared on all clinical parameters and no significant difference was found on any of them (p > .05 for all comparisons in Table 2).

Table 3.

Clinical data about the TDßT patients and patient subgroups.

Parameters	All patients	Patient subgroups
	TDßT (n = 62)	TM (n = 47)	TI (n = 15)	95% CI difference mean/median and/or p-value
Hb g/L
Median (IQR)	78 (17)	82 (18)	74 (8)	5.0 (−0.90 to 12.00)
ERBL × 109/L
Median (IQR)	19 (24)	14 (27)	23 (22)	5.9 (−4.00 to 14.00)
SF ng/mL
Median (IQR)	1000 (1945)	1200 (2250)	800 (880)	300 (−150 to 1054)
LIC mg/g d. w.
Median (IQR)	11.95 (14.30)	11.90 (14.60)	12.35 (11.10)	0.53 (−4.50 to 6.70)
Cardiac T2* ms
Median (IQR)	31.80 (11.50)	30.40 (16.30)	32.00 (12.50)	1.00 (−2.00 to 7.30)
Splenectomy
n (%)	45 (72.60)	36 (76.60)	9 (60)	p = .318
• Normopara-thyroidism n (%)	52 (83.80)	37 (78.70)	15 (100)	p = .100
• Hypopara-thyroidism n (%)	10 (16.20)	10 (21.30)	0 (0)
Fragility Fractures
n (%)	19 (30.6)	16 (34)	3 (20)	p = .356
Female	n = 30/62 (48.4%)	n = 24/47 (51%)	n = 6/15 (40%)
Eugonadism n (%)	15 (50)	11 (45.80)	4 (66.60)	p = .802
Hypogonadism with HRT n (%)	12 (40)	11 (45.80)	1 (16.70)	p = .263
Hypogonadism without HRT n (%)	3 (10)	2 (8.40)	1 (16.70)	p = 1.00
Male	n = 32/62 (51.6%)	n = 23/47 (49%)	n = 9/15 (60%)
Eugonadism n (%)	20 (62.5)	14 (61)	6 (67)	p = 0.532
Hypogonadism with HRT n (%)	5 (15.6)	4 (17)	1 (11)	p = .809
Hypogonadism without HRT n (%)	7 (21.9)	5 (22)	2 (22)	p = .785

IQR: interquartile range; Hb-haemoglobin; ERBL: Erythroblasts; SF: serum ferritin; LIC: liver iron concentration; Cardiac T2*-cardiovascular magnetic resonance T2*.

Analysis of serum sclerostin levels in the main groups and different subgroups

We assessed and compared serum sclerostin levels in the TDßT patients vs. the healthy controls and in different subgroups of the TDßT sample (Table 4). The results provided very strong evidence for a significantly higher sclerostin level in the TDßT patients (median 565.50 pmol/L) compared to the healthy controls (median 48.65 pmol/L), p < .001. The median difference of 517 pmol/L (95% CI: 365 pmol/L to 667 pmol/L) was of high magnitude corresponding to 11.62 times higher sclerostin level in the TDßT patients versus the healthy controls.

Table 4.

Comparisons of serum sclerostin levels in different subgroups of the TDßT.

Compared Groups	Median (IQR)	Median difference	95% CI median difference	p two-tailed
1.TDßT patients vs.	565.5 (592)
Healthy controls	48.6 (21)	517	365–667	< .001
2. TM patients vs.	649 (551)
TI patients	340 (534)	309	−59 to 358	.148
3. Male patients vs.	658 (772)
Female patients	466 (620)	192	−113 to 30	.602
4. Normoparathyroidism vs.	512 (600)
Hypoparathyroidism	660 (653)	149	−187 to 358	.730
	Mean (SD)	Mean difference	95% CI mean diff.	p two-tailed
5. Splenectomized vs.	702.7 (403)	287.60	80.00−495.00	.007
Non-splenectomized	415.1 (224)
6. Fragilityfractures vs.	829.9 (480)	307.55	109.00–506.00	.016
No fragility fractures	522.3 (290)

IQR-Interquartile Range; p ≤ .05 – ≤.01 weak evidence; p < .01 – >.001 strong evidence; p < .001 very strong evidence.

The comparison of serum sclerostin in splenectomized vs. non-splenectomized patients provided strong evidence for a significantly higher level in the splenectomized subgroup (mean level: 702.66 pmol/L ± 403) compared to the non-splenectomised (mean level: 415.07 ± 224 pmol/L), p = .007. The mean difference was 287.60 pmol/L (95%CI: 80 pmol/L to 495 pmol/L).

A significantly higher sclerostin level was also observed in the patients with fragility fracture events (mean 829.89 ± 480 pmol/L) vs. the patients who had not had a fracture event (mean 522.33 ± 290 pmol/L). The mean difference of 307.55 pmol/L, with 95% CI 109–506pmol/L, was supported by weak evidence (p = .016).

We did not find a significant effect related to the form of TDßT (TM vs. TI), p = 0.148; the patients’ sex (male vs. female), p =.602; and the type of parathyroidism (normoparathyroidism vs. hypoparathyroidism), p = 0.730.

We tested the diagnostic accuracy of serum sclerostin for identifying fragility fractures in patients with TDßT through ROC curve analysis (Figure 1). The results showed a significant, though not very high, diagnostic accuracy, AUC = 0.707 (CI: 0.558–0.857), supported by strong statistical evidence (p = .006). The associated criterion according to the Youden index indicated sclerostin >826 pmol/L (CI:791pmol/L–1046 pmol/L), with sensitivity = 60% and specificity = 85.71%. The patients with sclerostin levels >826 pmol/L were more likely to have fragility fracture events.

Figure 1.

ROC curve showing a fairly acceptable diagnostic ability of sclerostin in relation to fragility fracture events in TDßT patients.

Association of serum sclerostin with age, BMI, female and male gonadal state, Z-score at the lumbar spine (L1–L4), Z-score at the femoral neck and areal BMD in the lumbar spine and femoral neck

The potential relations between serum sclerostin and the following parameters were examined: age, BMI, female and male gonadal state (eugonadism, hypogonadism with HRT and hypogonadism without HRT), Z-score at the lumbar spine and femoral neck, and areal BMD in the lumbar spine and femoral neck.

Five significant associations were observed: a positive association with female gonadal state (r^s = 0.653, 95% CI:0.82–0.38, p < .001, N = 30); a negative association with Z-score at the lumbar spine (L1–L4) (r^s = −0.791, 95% CI: −0.86 to −0.67, N = 62, p < .001); a negative association with Z-score at the femoral neck (r^s = −0.708, 95%CI: −0.81 to −0.56, p < .001, N = 62); negative association with BMD at the lumbar spine (r^s = −0.773, 95% CI:−0.86 to − 0.65, N = 62, p < .001); negative association with BMD at the femoral neck (r^s = −0.629, 95% CI:−0.76 to −0.45, N = 62, p < .001). All five associations were supported by very strong statistical evidence.

Age, BMI and male gonadal state did not show a significant association with sclerostin levels: Age (r^s = 0.028, 95%CI: −0.22 to 0.27, N = 62, p = 0.791); BMI (r^s = −0.168, 95%CI: −0.14 to 0.34, N = 62, p = 0.109); male gonadal state (r^s = −0.124, 95%CI: −0.45 to 0.23, N = 32, p = 0.500).

The significant associations described above are illustrated on Figures 2–4. The individual sclerostin levels for each gonadal state of the female patients (eugonadism, hypogonadism with HRT and hypogonadism without HRT) and the mean values were plotted in order to illustrate the significant relationship. The highest individual levels and mean value were observed in the female patients with hypogonadism without HRT and the lowest in the eugonadic female patients (Figure 2).

Figure 2.

Individual value plot of sclerostin levels vs. gonadal state illustrates the significant association between the two parameters. The highest individual levels and mean value were observed in the female patients with hypogonadism without HRT; and the lowest in the eugonadic female patients.

Figure 3.

Panel A – negative association between sclerostin levels and Z-scores at the lumbar spine. Panel B – negative association between sclerostin levels and Z-scores at the femoral neck. Panel C – negative association between sclerostin levels and areal BMD (g/cm²) in the lumbar spine. Panel D – negative association between sclerostin levels and areal BMD (g/cm²) in the femoral neck.

Figure 4.

Panel A – negative association between sclerostin and Hb levels. Panel B – positive association between sclerostin and LIC.

The significant negative association of sclerostin levels with Z-scores and BMD (g/cm²) at the lumbar spine (L1–L4) and femoral neck are illustrated on Figure 3. Lower Z-scores and BMD (g/cm²) are associated with higher sclerostin levels.

Association of serum sclerostin with thalassaemia specific parameters: haemoglobin (Hb), serum ferritin (SF), liver iron concentration (LIC) and cardiac T2*

Further correlation analyses explored the connection between serum sclerostin levels and thalassaemia specific parameters, including Hb, SF, LIC, and cardiac T2*. Two significant relationships were found: a negative association with Hb (r^s = − 0.423, 95%CI −0.60 to −0.19, N = 62, p < .001); and a positive association with LIC (r^s = 0.367, 95%CI: 0.13–0.56, N = 62, p = .003). The other two parameters, SF and cardiac T2*, did not show a significant association with serum sclerostin (SF r^s = 0.058, 95%CI: −0.19 to 0.30, N = 62, p = 0.653; Cardiac T2* r^s = −0.198, 95%CI:−0.43 to 0.05, N = 62, p = 0.127). Figure 4 shows the significant association of sclerostin levels with Hb and LIC. Higher sclerostin levels are associated with lower Hb levels and with higher LIC values.

Association of serum sclerostin with markers of bone metabolism (Beta-CTx, OCN, sRANKL and OPG)

Sclerostin levels showed a significant relationship with all examined bone metabolism markers, including: a positive association with Beta-CTx (r^s = 0.890, 95% CI: 0.82–0.93, N = 62, p < .001), a negative association with OCN (r^s = −0.619, 95% CI: −0.75 to −0.44, N = 62, p < .001), a positive association with sRANKL (r^s = 0.323, 95% CI: 0.08–0.5, N = 62, p = .002), and a positive association with OPG (r^s = 0.320, 95% CI: 0.08–0.5, N = 62, p = .002). These relationships are illustrated on Figure 5.

Figure 5.

Panel A – positive association of sclerostin with Beta-CTx; Panel B – negative association with OCN; Panel C – positive association withsRANKL; Panel D – positive association with OPG.

Discussion

As mentioned earlier, research about serum sclerostin levels in beta-thalassaemia patients is scarce. The few existing studies have involved only osteoporotic thalassaemia cases [8,9,38,39]. Voskaridou et al. and Morabito et al. reported a moderate positive correlation between the levels of sclerostin in serum and BMD in osteoporotic thalassaemia patients [8–9]. Moinzadeh et al. found a significant relation between the sclerostin serum levels and pretransfusion haemoglobin in patients with TM [39]. What our study adds to the existing research is a multilayered account of the role of serum sclerostin in transfusion-dependent beta thalassaemia patients, including both TM and TI cases, in relation to a range of thalassaemia-specific parameters and markers of bone metabolism.

First, our comparison of serum sclerostin in transfusion-dependent beta thalassaemia patients and healthy controls showed 11.62 times higher levels in the patient group.

Based on the very strong statistical evidence which we obtained in the comparison of TDßT patients with healthy controls, it appears that elevated sclerostin levels are characteristic of transfusion-dependent beta thalassaemia patients. Based on the 95% CI, we estimated that serum sclerostin levels could be between 7 and 13 times higher in replications with similar populations. However, it should be mentioned here that our results are in contrast to those of Tsartsalis’ study which showed no significant difference in the sclerostin levels between osteoporotic TM patients and healthy controls [38]. We consider that a plausible explanation for this discrepancy is due to an important difference between our study and Tsartsalis’. Whereas our sample included patients naïve to osteoporotic treatment, Tsartsalis’ involved patients who were on osteoporotic treatment. In their conclusions the authors mention that the lack of significant differences between TM patients, osteoporotic patients and healthy controls regarding serum sclerostin and CTX levels may be attributable to the effectiveness of the present-day osteoporotic treatment.

Within the patient group, the TM and TI subgroups did not differ significantly in circulating serum sclerostin (p = 0.148, 95% CI of the median difference: −59 pmol/L to 358 pmol/L). This finding needs further validation since we were unable to triangulate our results with previous reports due to the lack of related research, comparing TM and TI patients on serum sclerostin levels.

Second, we examined the potential diagnostic ability of sclerostin in relation to fragility fractures, which according to the World Health Organizatin define the patients with severe osteoporosis [40]. Moreover, it has been reported that patients who have already experienced fracture events have an increased risk of subsequent fractures [41]. Our results showed significantly higher levels of sclerostin in patients with experienced fragility fractures as compared to patients without fractures with a median difference of 308 pmol/L (95%CI: 109 pmol/L to 506 pmol/L). These data prompted a follow-up ROC curve analysis about the diagnostic ability of sclerostin regarding fragility fracture events. We established an acceptable diagnostic ability of sclerostin for distinguishing patients with fragility fracture events from those without fractures, with 60% sensitivity and 85.71% specificity associated with a criterion value > 826 pmol/L. Extrapolating from these results, it appears that sclerostin has the potential to serve as one of the markers associated with severe osteoporosis in beta thalassaemia patients. However, this claim needs futher validation considering the relatively large 95%CI (791 pmol/L–1046pmol) and the lack of previous reports concerning sclerostin as an indicator of fragility fracture events (i.e. severe osteoporosis) in TDßT.

Third, possible associations of serum sclerostin with a range of BMD parameters were examined in our study, yielding a significant negative trend between Z-scores and areal BMD (g/cm²) at the lumbar spine and the femoral neck. These results add further evidence about the potential role of circulating sclerostin as a marker of bone damage in patients with TDßT. However, our results are not in line with the findings of Voskaridou [8] and Morabito et al. [9] who reported a positive relationship in thalassaemia patients with osteoporosis. They are also different from Moinzadeh’s study [39], which showed no relation between sclerostin serum levels and bone mineral density in osteoporotic TM patients. We presume that the difference could be due to the fact that in our study the patient group included not only osteoporotic, but also osteopenic and TDßT patients with normal BMD, thus respresenting a wider range of patient characteristics vs. the aforementioned studies which explored the same association in samples that were more homogeneous (restricted range).

Fourth, we analysed the potential associations of serum sclerostin with a range of thalassaemia-specific parameters (Hb, SF, LIC, and cardiac T2) and with markers of bone metabolism (OCN, Beta-CTx, OPG and sRANKL). From the thalassaemia-specific parameters, two showed significant associations with serum sclerostin. A negative association was established with pretransfusion haemoglobin and a positive with the severity of liver iron overload. The negative association with pretransfusion haemoglobin was supported by very strong statistical evidence, however the 95%CI showed a rather large range of coefficients between −0.60 and −0.19, cautioning against any hasty conclusions. The uncertainty is coupled by the lack of related research specific to thalassaemia patients (at least to our knowledge), except for that by Moinzadeh et al. [39] Yet, our finding can not be compared with the latter study because Moinzadeh et al. did not explore the strength and direction of the relation between sclerostin and pretransfusion haemoglobin.

The relationship of sclerostin with liver iron overload was significant, however not very strong. Yet, we consider that this positive association may reflect the close link between the hepatic and bone marrow iron overload. Related research suggests that this phenomenon may partially be due to a reduced hepatic metabolism of sclerostin in the case of iron-induced hepatic dysfunction [42]. On the other hand, recent research by Ehnert et al. [43] and other teams has shown that deseased liver tissue from different ethiology expresses higher sclerostin levels [44–47]. Extrapolating from their findings, it may also be hypothesised that iron loaded liver tissue may actively synthesise sclerostin.

Furthermore, the observed negative association between OCN and sclerostin in our study collaborates previous findings about the role of sclerostin as a Wnt-inhibitor of osteoblast synthesis in clinical and in-vitro settings [2,5,6,13,14]. In this line of research, the low levels of OCN in our beta-thalassaemia patients can be related to a decreased neoformation phase [48] as a response to elevated sclerostin levels [49].

Another of our findings is the observed difference in sclerostin levels between splenectomized and non-splenectomized patients. The sclerostin level in the splenectomized subgroup was higher by 287.60 pmol/L. The interpretation of this result is speculative given the relatively large 95% CI of the mean difference (80 pmol/L to 495 pmol/L) and the lack of previous research about the role of the spleen on sclerostin levels. Further research is needed in order to determine whether the spleen exerts a primary role on sclerostin methabolism through modulating the functions of the lymphocytic populations or other cells of the immune system, or a secondary through iron redistribution after splenectomy.

Our study also examined the relationship between serum sclerostin and gonadal state (eugonadism, hypogonadism with HRT and hypogonadism without HRT) separately for the male and female beta-thalassaemia patients. Different trends were revealed in the two subgroups. In the female patients, sclerostin showed a strong positive association with gonadal state. The highest mean and individual levels were associated with the female patients with hypogonadism without HRT and the lowest with the eugonadic female patients. Our findings are in line with experimental models, demonstrating that SOST-gene expression is suppressed by 17-OH oestradiol [50–53]. It must also be mentioned here that the 95% CI showed a rather large range between the lowest (r^s = 0.38) and the highest (r^s = 0.82) expected values of the correlation coefficients, but this in our opinion should be attributed to the small sample size (N = 30) of female patients. With a bigger sample, the level of uncertainty/variability will be reduced. In the male subgroup, no significant association was found between sclerostin and gonadal state. This result is not surprising when we consider the findings of Mödder et al. [51] which show that in adult men with age-related testosterone deficiency, sclerostin levels were not influenced by testosterone supplementation [50]. Thus, it can be speculated that the biological role of the sex hormones on the SOST-gene expression may follow different pathways in both sexes [14,53].

Another hypothezied relation between sclerostin levels and the type of parathyroidism (normoparathyroidism vs. hypoparathyroidism) did not reveal significant differencies in the sclerostin levels associated with the two types of parathyroidism. However, our findings can only be compared with those of studies involving other physical conditions due to the lack of related research with thalassaemia patients. For instance, our results do not corroborate the findings reported in Mirza [54] and Bhattacharyya [55]. Both studies analysed sclerostin levels and parathyroid function in postmenopausal women, and both found that serum sclerostin levels correlated with parathyroid activity. Considering the lack of research with thalassaemia patients and the contrasting findings reported in relation to postmenopausal women, further research, and with larger samples, is necessary in order to validate the observed lack of significant association between sclerostin levels and the type of parathyroidism in thalassaemia patients.

Extrapolating from our results about age and BMI, which did not show a significant association with sclerostin levels, it can be presumed that these factors exert an inferior influence on sclerostin methabolism in thalassaemia patients. A similar trend was observed in studies with postmenopausal osteoporotic women [29].

Regarding OPG and sRANKL, our study adds further evidence about the close interplay between the Wnt-signalling pathway and the OPG/RANKL system [56,57]. Although the significant positive association of serum sclerostin with sRANKL was expected, the significant positive association with OPG levels requires more careful interpretation. On the one hand, the elevated OPG levels in our TDßTgroup corroborate the findings reported by Pietrapertosa et al. [58] in a study with osteoporotic beta-thalassaemia patients. On the other hand, our results are contradictory to the conclusions of in-vitro and animal studies which have shown that OPG plays a protective role in bone methabolism [59]. To cast some light on this contradiction, we draw on the findings of clinical studies with different patient populations [60,61] whose findings suggest that the protective role of OPG in bone metabolism may not be as straigforward as generally thought. For instance, Sherief et al. observed increased OPG levels in beta-thalassaemia patients in association with their atherosclerotic changes [62]. Besides, the high serum levels of OPG in our patient group may partially be attributed to the exsistence of cardiovascular damage, another significant and closely related to osteoporosis comorbidity in thalassaemia patients [63]. Furthermore, elevated sclerostin and OPG levels have been associated with an inflamatory states [64]. Thus, our results may reflect a systemic, multi-tissue synthesis due to the pro-inflamatory profile of the beta–thalassaemia disease [65,66]. This link with inflammation could account not only for the elevation in sclerostin and OPG levels, but also for the positive association between them. This line of reasoning may appear somewhat contradictory to Voskaridou’s hypothesis that sclerostin is probably icreased in TDßT due to osteocytic overgrowth [8]. We consider that one way to clarify some of the existing controvercies is through a parallel evaluation of sclerostin levels in serum and its expression in bone and other tissues. Regardless of the source and mechanism of its production, sclerostin strongly interferes with the degree of bone damage in patients with TDßT. The significant associations, revealed by our data, of sclerostin levels with B-Ctx and sRANKL further confirm its detrimental effect on bone health.

Of course, we must recognise that sclerostin is not the only molecule that affects the Wnt-signalling pathway. Dkk-1- is another known Wnt-signalling inhibitor, whose levels are increased in thalassaemia patients. Detailed information is provided by several treatment-related studies by Voskaridou et al. [8,67,68]. In the earlier study, Voskaridou et al. observed a negative association between Dkk-1 levels and low bone mineral density in thalassaemia patients with osteoporosis [67]. In the same study, Dkk-1 levels decreased after threatment with zolendronic acid. However, in the subsequent studies which involved sclerostin, the treatment with zolendronic acid or denozumab did not reduce the levels of sclerostin [8,68]. In a study of another anti-osteoporotic drug, strontium ranelate, Morabito et al. evaluated both serum sclerostin and Dkk-1 pre-and 24 months after treatment. The results showed no significant change in the Dkk-1 levels post-treatment, whereas the levels of sclerostin decreased [9]. We referred to the above findings from treatment studies in order to make an important point. Although our study focussed on sclerostin as one of the known molecules which affects the Wnt-signalling pathway, we recognise that effective diagnosis, treatment and monitoring of thalassaemia-related osteoporosis can not be achieved based solely on a patient’s sclerostin profile. On the contrary, clinicians should take into consideration multiple parameters and factors. Sclerostin is one of them which according to our results can help a clinician’s choice of an appropriate treatment for each individual case.

Strengths and limitations of the study

The main strength of our study is that it has extended the existing body of research about serum sclerostin to a full spectrum of transfusion-dependent beta thalassaemia patients with normal BMD, decreased BMD and osteoporosis, and has revealed its association with thalassaemia-specific and bone disease parameters. Alongside this, the following limitations need to be mentioned: (1) Sclerostin levels in serum were measured once. Tracing them for a longer period of time in parallel with BMD parameters and markers of bone metabolism is likely to provide more precise information about its possible predictive value for future bone loss; (2) In analysing the association of sclerostin with various characteristics of the transfusion-dependent beta thalassaemia patients, we used subgroups of relatively small size.

Our results need to be validated by further research with larger sample sizes that assure adequate power; 3) Considering the fact that thalassaemia patients represent a model of multiple tissue alterations, we recognise the need for further research about the relationship of sclerostin burden with a variety of tissue damages and organ dysfunctions in beta thalassaemia patients.

Conclusion

In the introduction of this article, we outlined the unique nature of thalassaemia-related bone disease and the need for empirical evidence that will help understand the complex multitude of parameters associated with this condition. In our study, we focussed on the role of serum sclerostin as one of the many parameters that we hypothesised could play a role in the diagnosis, treatment and prevention of osteoporosis in TDßT patients. In sync with our goal, we have provided a comprehensive account of clinically relevant associations of sclerosin serum levels with bone mineral density; bone synthesis and resorption markers; and thalassaemia-related alterations in TDßT. Extrapolating from our results, we have identified the following main trends:

1. Adult patients with TDßT express abnormally high serum levels of sclerostin which are negatively associated with bone mineral density and with bone synthesis markers and positively associated with bone resorption markers.

2. Some of the thalassaemia-related factors which appear to be assotiated with the elevated serum sclerostin levels in TDßT patients include: pretransfusion levels of haemoglobin, liver iron burden, splenectomy status, fragility fracture events and female hypogonadism.

3. Age, sex, BMI, cardiac iron overload, severity of thalassaemia disease, male hypogonadism as well as hypoparathyroidism do not seem to influence the level of circulating sclerostin in this patient category.

On the basis of these results, we conclude that sclerostin plays a role in the bone pathophysiology of beta-thalassaemia patients and propose that it may serve as a marker of severe osteoporosis, whose clinical manifestation involves fragility fractures in this patient population. Moreover, sclerostin can be a potential target in the prevention of severe bone conditions related to beta-thalassaemia.

Funding Statement

This study was financially supported under Project No 02/2017 of the Medical University, Plovdiv, Bulgaria.

Disclosure statement

No potential conflict of interest was reported by the authors.

Author Contributions

KS developed the study design, literature search, variables extraction and statistical processing of the data and wrote the manuscript of the article.VG-M performed literature search. PG collected the data from the patient’s medical records and performed the process of patients’ selection and signing the written informed concent. TD performed all the laboratory tests of the studied population and ST performed the DXA measurments. All the authors have read, edited and approved the manuscript. ZhG approved the final version of the manuscript. KS takes responsibility for the integrity of the data analysis.