Relationship between ectopic calcifications and bone fragility depicted on computed tomography scan in 70 patients with systemic sclerosis

Abstract

Background:

A higher risk of osteoporotic fracture was described in systemic sclerosis patients than in healthy patients.

Objective:

To evaluate the relation between osteoporotic fracture risk measured by the scanographic bone attenuation coefficient of the first lumbar vertebra (SBAC-L1) on computed tomography (CT) scan and the presence of ectopic calcifications: vascular, valvular and spinal.

Methods:

This monocentric retrospective study was performed on patients followed between 2000 and 2014 at Nancy University Hospital. Systemic sclerosis patients, according to ACR/EULAR 2013 criteria, followed from 2000 to 2014 and who underwent, during their follow-up, a CT including the first lumbar vertebra were included. The SBAC-L1 was measured with a threshold set at 145 Hounsfield units (HU). Vascular and spinal calcifications were studied on CT. For vascular calcifications, the Agatston score was used. Valvular calcifications were studied on echocardiography.

Results:

A total of 70 patients were included (mean age: 62.3 (±15.6) years, women 88.5%). The mean SBAC-L1 was 157.26 (±52.1) HU, and 35 patients (50%) presented an SBAC-L1 ⩽ 145 HU. The reproducibility of the calcification evaluation was good, with kappa coefficients varying between 0.63 and 1. In univariate analysis, spinal and vascular calcifications were associated with an SBAC-L1 ⩽ 145 HU, with ORs of 13.6 (1.6–113.3) and 8 (95%CI: 2.5–25.5), respectively. In multivariate analysis, the SBAC-L1 was not associated with the presence of any ectopic calcifications. The SBAC-L1 decreased with age (p = 0.0001).

Conclusion:

Patients with systemic sclerosis with an SBAC-L1 ⩽ 145 HU were older, but they did not have more ectopic calcification.

Trial registration:

The ethics committee of Nancy Hospital agreed with this study (referral file number 166). This study was designed in accordance with the general ethical principles outlined in the Declaration of Helsinki.

Background

Systemic sclerosis is a disease characterised by ectopic calcifications on the hands (periarticular calcifications), ¹ on vessels (vascular calcifications), or in the heart (valvular calcifications).^2,3 Prognostic factors depend on pulmonary, cardiac or renal involvement. Life expectancy has improved in recent years,^{4 –7} which is why osteoporosis has to be screened in a population with a large predominance of postmenopausal women. ⁶ The prevalence of osteoporosis assessed by dual-energy X-ray absorptiometry (DXA) in sclerodermic patients varies from 3% to 60%^{8 –10} and is still debated when compared to the general population.^{8,9,11 –24} The risk of osteoporotic fracture is higher in this population, with a relative risk of 1.78 for vertebral fracture and 1.86 for hip fracture compared to healthy paired patients. ²⁵ The pathophysiological mechanisms of osteoporosis in systemic sclerosis are not well identified and are related to inflammation, malabsorption and vitamin D deficiency^{9,13,14,17,21,23,24} and/or to a loss of mobility in severe disease.^{9,13,14,17,21,22,23,24} Some authors also have proposed specific pathophysiological mechanisms with the displacement of calcium stocked in bone towards vascular and soft tissues (which is responsible for calcinosis) and around joints (hands and feet) as well as ectopic calcifications on the spine, muscles, ligaments and bursa.^14,21 This hypothesis is also supported by a previous study demonstrating a relationship between calcic deposits on the hands of systemic sclerosis patients and low scanographic bone attenuation coefficient of the first lumbar vertebra (SBAC-L1) values on CT. ²⁶

Usually, osteoporosis screening is based on dual-energy X-ray absorptiometry (DXA), considered the gold standard method. ²⁷ However, in systemic sclerosis patients, ectopic calcifications could create an overestimation of the bone mineral density (BMD) measured on DXA.^{28 –32} Moreover, osteoporosis screening is often forgotten during the clinical follow-up of systemic sclerosis patients.

Computed tomography (CT) scans are performed in systemic sclerosis patients to assess systemic involvement (lung) or to explore adverse events related to immunosuppressive treatment. CT presents a double advantage: first, it permits a vertebral fracture prevalence evaluation on sagittal reconstructions, and second, it permits a trabecular bone density evaluation through the scanographic bone attenuation coefficient of the first lumbar vertebra (SBAC-L1). ³³ An SBAC-L1 value ⩽ 145 Hounsfield units (HU) is usually considered a threshold for a vertebral fracture (VF) since Pickhardt et al. showed that this value depicted 96.6% of patients with VFs in general the population, whereas DXA (with a T-score ⩽ –2.5 standard deviation (SD)) identified only 39% of these patients. Moreover, CT also offers the ability to assess calcic deposits in large vessels and to determine cardiovascular risk although the Agatston et al. ³⁴ and Mori et al. ³⁵ score. This calcium scoring is a validated evaluation method and has been shown for being correlated with a higher cardiovascular risk. ³⁶ Therefore, this calcium scoring may provide important prognostic information. ³⁷ To our knowledge, no study has evaluated the relationship between bone fragility measured by SBAC–L1 and cardiovascular risk assessed by the Agatston score on CT scan.

We hypothesised that both vascular and spinal calcic deposits, reflecting risk factors for systemic sclerosis severity, may be related to a bone fragility process assessed by the SBAC-L1.

Thus, the objective of this study was to evaluate the relation between the SBAC-L1 and the presence of the following ectopic calcifications: aortic vascular, valvular or spinal calcifications.

Methods

Demographic and clinical data

This descriptive analytical and retrospective study was conducted in patients who were followed at Nancy University Hospital (France) between January 2000 and April 2014 for diffuse or limited systemic sclerosis. We chose medical records with codes M34.0, M34.1, M34.8 and M34.9 of the International Classification of Diseases and corresponding to the classification of ACR/EULAR 2013 for systemic sclerosis. ³⁸ Patients were included if they had undergone a thoracic or lumbar or thoraco–abdomino–pelvic (TAP) CT during the follow-up.

Demographic (age, sex and disease duration), clinical (medical history, arthralgia, modified Rodnan score and Raynaud syndrome) and biological (C-reactive protein (CRP), erythrocyte sedimentation rate, antinuclear antibodies (ANA) and phosphocalcic evaluation) characteristics and treatments concerning systemic sclerosis were collected from all of the medical records (before the CT). The selected complementary exams (radiographs of the hands, pulmonary function testing (PFT) and cardiac echography) had to be conducted in the year before or after CT evaluation. Hand structural lesions were studied on the radiographs of the hands and according to Erre et al. ³⁹ Score by a senior rheumatologist with much experience in this scoring method. Pulmonary involvement was studied functionally with PFT and morphologically with thoracic or TAP CT (non-specific interstitial pneumonia (NSIP), usual interstitial pneumonia (UIP) and interstitial lung disease (ILD)) by a senior radiologist. Pulmonary hypertension (PH) was detected on cardiac echography and confirmed by right heart catheterization. ⁴⁰

For the assessment of osteoporosis, clinical risk factors (sex, age, chronic biological inflammation, smoking and corticosteroid therapy), bone density on DXA and anti-osteoporotic treatments were recorded during the entire follow-up. A comparison between DXA and SBAC-L1 measurements was performed if the delay between both exams was less than 2 years. For DXA, osteoporosis is classically defined in patients with a T-Score ⩽ –2.5 SD on the spine, and osteopenia is defined in those with −2.5 SD < T-score ⩽ –1 SD.

Ectopic calcification evaluation: aortic vascular, aortic or mitral valvular, and spinal

Aortic vascular calcifications were diagnosed on CT using the calcium scoring method in Osirix software, according to the Agatston score, on axial reconstructions with a thickness equal to 3 mm. The density of the vascular lesions ranged from 130 to more than 400 HU. ³⁴ The threshold for a vascular calcific lesion was set at a computed tomographic density of 130 HU for a minimum area (⩾1 mm²). This eliminated single pixels with a computed tomographic density > 130 HU due to noise.

The aorta, from the brachiocephalic artery to its terminal division, was divided into five segments: segment 0, the ascending aorta, from the aortic valve to the brachiocephalic artery; segment 1, the horizontal aorta, between the brachiocephalic artery and the aortic arch; segment 2, the descending aorta, between the aortic arch and the diaphragm pillar; segment 3, the visceral aorta, from the diaphragm pillar to the renal artery; and segment 4, the terminal aorta, from the renal artery until its division (Figure 1). The ascending aorta (segment 0) was not included in this study because the aortic arch and the proximal descending aorta contained most of the calcifications. ⁴¹ The terminal aorta was viewable only on TAP CT. First, the volume of the vascular calcifications was calculated for each segment. A region of interest (ROI) was drawn around all vascular calcifications for each slice of the corresponding segment adjusted if necessary by manual drawing and expressed in square millimetres. For each aortic segment, the volume of the vascular calcifications was calculated as the sum of the area of the vascular calcification present in the slices of the corresponding segment and expressed in mm³. Second, a binary evaluation was performed according to the presence or absence of calcification for the four aortic segments.

Figure 1.

Aortic vascular calcification evaluation: (a) aorta division in five parts: 0 – the ascending aorta, from the aortic valve to the brachiocephalic artery (not studied); 1 – the horizontal aorta, between the brachiocephalic artery and the aortic arch; 2 – the descending aorta, between the aortic arch and the diaphragm pillar; 3 – the visceral aorta, from the diaphragm pillar to the renal artery; and 4 – the terminal aorta, from the renal artery until its division. (b) Aortic vascular calcium scoring on CT axial sections in parts 1, 2 and 3: the region of interest (in red, blue and yellow, according the studied segment) was drawn around all calcic lesion; automated measurements of the lesion area in square millimetres and the maximal computed tomographic number of each region of interest were recorded.

Valvular calcifications (aortic and mitral) were binary studied on echocardiography by a senior cardiologist.

Spinal calcifications were diagnosed by a senior rheumatologist, blinded to the patient’s characteristics, on CT spinal axial sections in bone windows. Sagittal and coronal reconstructions were performed with the aim of obtaining a multiplanar assessment. Each discovertebral level was scored for the presence of calcification and its topography (anterior: anterior, central or posterior discovertebral space; intracanal; posterior: facet joints, costotransverse and interspinous) (Figure 2). Calcification was defined if it was found on at least two contiguous slices and confirmed in one of the two other planes.

Figure 2.

CT spinal calcification evaluation. Spinal calcification evaluation on CT in axial sections and sagittal reconstructions: (a, b) Anterior calcifications in axial sections. Intracanalar calcifications in axial section (c) and sagittal reconstruction (d). (e), (f) and (g): Posterior calcifications: Pseudo-tumoral calcifications with posterior inter-apophyseal joint deposits in axial sections (e, f). Pseudo-tumoral peri-spinous calcification in sagittal reconstruction (g).

For the CT calcification evaluation (vascular and spinal calcifications), a second interpretation was performed by a senior rheumatologist to assess the inter-reader reproducibility. The first reader made two evaluations to assess intra-reader reproducibility.

Vertebral fracture assessment and the densitometric scanographic bone attenuation coefficient of L1 (SBAC-L1)

All CT were conducted in the same hospital and 43 TAP, 26 thoracic and 1 lumbar CT were selected. From the axial acquisitions, sagittal reconstructions in the bone window permitted a vertebral morphology study from C7 to L1 for thoracic CT and from C7 to S1 for TAP CT based on OSIRIX software (v6.5.1–64 bits). The VF was analysed according to an adaptation of Genant et al.’s ⁴² classification, which is typically used on spine radiographs. The grade of the VF was determined on the most severe lesion observed on sagittal sections.

The SBAC-L1 study was conducted on L1 axial sections through the pedicles on the bone window. The largest elliptical ROI was drawn in trabecular bone and provided the average bone mineral density in HU, with excellent intra- and inter-reader reliabilities. ²⁶ We used the mean of the two intra-reader values. The threshold of 145 HU defined by Pickhardt et al. ³³ was used as the fracture threshold to classify patients at risk of a VF.

Ethical approval and consent to participate

All data were available from usual care in patients with systemic sclerosis. The ethics committee of Nancy Hospital agreed with this study (referral file number 166). This study was designed in accordance with the general ethical principles outlined in the Declaration of Helsinki. Patients gave their consent to use their medical data when they were cared for at the university hospital.

Statistical analysis

Both descriptive and comparative analyses were conducted by accounting for the nature and distribution of the variables (normality assessed by Kolmogorov–Smirnov test). Qualitative variables were described with frequencies and percentages; quantitative variables were evaluated with the mean ± SD (standard deviation) or with the median and interquartile range (IQR).

The Kolmogorov–Smirnov test showed that among the continuous demographic and clinical variables, only age followed a normal distribution. Student’s t-test was used for age, and Mann–Whitney U test was used for the other variables. For qualitative variables, the chi-square test and/or the exact calculation of Fisher was used. A logistic regression was performed to test variables significantly associated with the binary outcome SBAC-L1 ⩽ 145 HU. Significant results (univariate and multivariate analysis) are presented with the odds ratio (OR) and its 95% confidence interval (95% CI). To analyse the intra-reader and inter-reader reliability, we used Cohen’s kappa method. The α coefficient was established as 0.05, except for the inter-reader reliability study, where it was established at 0.01 given the repetition of the tests. IBM™ SPSS Statistics v23 was used for the data analysis.

Results

Demographic and clinical data

One hundred patients were selected from our review of medical records: 70 of them met the ACR/EULAR 2013 criteria for systemic sclerosis, and all of them benefitted from CT. The population characteristics are presented in Table 1. Sixteen patients (22.9%) had diffuse disease, and 54 (77.1%) had limited cutaneous disease. Seventeen patients (24.3%) had hypertension, 7 (10%) had diabetes, 29 (41.4%) had dyslipidaemia and 9 (12.9%) were obese. Concerning the known clinical risk factors for osteoporosis, 47 women (67.1%) were menopausal (>50 years), 17 (24.3%) patients smoked, 20 (28.6%) presented biological inflammation and 28 (40%) received corticosteroid treatment (average posology: 3.8 ± 6.6 mg/day). Sixty patients (85.7%) had at least one clinical or biological risk factor for osteoporosis. Eighteen patients (25.7%) received vitamin calcic supplementation and 10 (14.3%) received a specific treatment for osteoporosis (antiresorptive therapy).

Table 1.

Characteristics of patients.

	N = 70
Demographic/clinical data
Age (SD, years)	62.4 ± 15.6
Disease duration	10 (5–13)
Women	62 (88.6)
Smoking	17 (24.3)
Digital ulcers	36 (51.4)
Capillaroscopy (n = 52)	46 (65.7)
Arthralgia	58 (82.9)
Rodnan score	6 (2–11)
Dyspnoea stage III–IV	28 (40)
Death	9 (12.9)
Dyslipidaemia	29 (41.4)
Obesity (BMI > 30 kg/m2)	9 (12.9)
Gastroesophageal reflux	42 (60)
Biology
ESR (mm)	16.5 (10–32)
C-reactive protein (mg/L)	4.4 (1.6–7.1)
Anticentromere antibodies (n = 63)	38 (54.3)
Anti-SCL-70 antibodies (n = 63)	11 (15.7)
Radiography and CT
Calcinosis	16 (22.9)
Periarticular calcifications	20 (28.6)
Acro-osteolysis	14 (20)
Pulmonary involvement
NSIP	9 (12.9)
UIP	5 (7.1)
Interstitial lung disease	18 (25.7)
DLCO/VA (n = 66)	18 (25.7)
PFT restrictive respiratory disorder (n = 66)	6 (8.6)
PFT obstructive respiratory disorder (n = 66)	14 (20)
PH (n = 66)	8 (11.4)
Treatment
Corticosteroid therapy	28 (40)
DMARDs	19 (27.1)
Immunosuppressive drugs	6 (8.6)
Osteoporosis ⩾ 1
Clinical risk factor	60 (85.7)
Number of DXA	30 (42.8)
Osteoporosis on DXA	5 (7.1)
VF on CT	3 (4.3)
SBAC-L1 (HU)	145.6 (120.9–189)
Vitamin calcic supplementation (n = 40)	18 (25.7)
Specific treatment for osteoporosis (n = 40)	10 (14.3)
Spinal calcifications
All locations	59 (84.3)
Anterior	57 (81.4)
Posterior	24 (34.3)
Intracanalar	22 (31.3)
Foraminal	8 (11.4)
Aortic vascular calcifications	45 (64.3)
Valvular calcifications
Aortic (N = 61)	21 (30)
Mitral (N = 61)	18 (25.7)

Qualitative data are described by the count (percentage), whereas quantitative data are described by the average ± SD for age or by median and interquartile index. Pulmonary involvement was determined by a specific CT evaluation for interstitial lung disease (ILD) in two categories: usual interstitial pneumonia (UIP) and non-specific interstitial pneumonia (NSIP). The duration of disease progression corresponded to the time between the date of diagnosis and the date of the imaging examinations.

BMI: body mass index; CT: computed tomography; DLCO/VA: CO lung diffusion capacity; DMARD: disease-modifying antirheumatic drug; DXA: dual-energy X-ray absorptiometry; ESR: erythrocyte sedimentation rate; PFT: pulmonary function testing; PH: pulmonary hypertension; SBAC-L1: scanographic bone attenuation coefficient of the first lumbar vertebra; SD: standard deviation; VF: vertebral fracture; HU: Hounsfield units.

Vertebral fracture assessment and the scanographic bone attenuation coefficient of L1 (SBAC-L1)

For CT, three women presented with a vertebral fracture (4.3%). Two grade 2 and one grade 3, according Genant et al.’s ⁴² classification.

The mean SBAC-L1 was 157.3 ± 52.2 HU, and 35 patients (50%) had an SBAC-L1 ⩽ 145 HU. The mean of the two intra-reader values was used to calculate the SBAC-L1. The SBAC-L1 values of the three patients with VFs were 102.5, 131.4 and 171 HU. There was no difference in SBAC-L1 between patients with diffuse or limited systemic sclerosis (p = 0.3). ²⁶

Thirty patients (42.8%) underwent DXA during their follow-up: five (16.7%) were diagnosed with osteoporosis. None of these osteoporotic patients had a VF.

Ectopic calcifications assessment

Intra- and inter-reader reliabilities

Concerning vascular calcification, inter- and intra-reader reproducibilities were very good, with kappa coefficients varying between 0.81 and 1.00 according to the studied segment of the aorta.

Concerning spinal calcifications, inter- and intra-reader reproducibilities were good to very good, with kappa coefficients ranging from 0.63 to 0.89 according to the calcification location. For the binary evaluation of spinal calcification (all locations combined), kappa was 0.714 [0.001 to 1] and 0.63 [–0.047 to 1] for inter- and intra-reader reliabilities, respectively.

Vascular and valvular calcification assessment

For aortic vascular calcifications, 45 patients (64.3%) were affected, with a mean volume of 1839.4 mm³ (±3499.2). For segment 2 (horizontal aorta), 51.4% of the patients had calcium deposits with a mean volume of 472.7 mm³ (±1240.8). For segment 3 (aorta), 64.3% of the patients had calcium deposits with a mean volume of 288.9 mm³ (±533.1). The calcium deposit in segment 4 (visceral aorta) could be evaluated in 61 patients, and 68.9% of them had a mean volume of 330 mm³ (±672.3). For segment 5 (terminal aorta), 50 patients were evaluated, and 68% of them had calcium deposits with a mean volume of 1121.4 mm³ (±2028.8).

Finally, for valvular calcifications, 21 patients (30%) and 18 patients (25.7%) had aortic or mitral valvular calcifications, respectively.

There was no difference between the patients with diffuse or limited disease for aortic vascular, aortic or mitral valvular or spinal calcifications (p = 0.051, 0.162, 0.167 and 0.052, respectively).

Spinal calcification assessment

Of the 70 patients, 59 (84.3%) had at least one spine calcification; 57 (81.45%) had lesions of the anterior segment. Intracanal lesions were observed in 22 patients (31.3%), including 8 (11.4%) foraminal localisations with a pseudo-tumoral appearance. Twenty-four patients (34.3%) had posterior spinal lesions.

There was no difference between the patients with diffuse or limited disease for ectopic calcifications (aortic vascular, aortic or mitral valvular or spinal).

Specific risk factors associated with an SBAC-L1 ⩽ 145 HU

In univariate analysis (Table 2), age was significantly associated with an SBAC-L1 ⩽ 145 HU (OR = 1.09 (95% CI: 1.40–1.13), p = 0.0001). For spinal calcifications, only anterior localisation was associated with an SBAC-L1 ⩽ 145 HU (OR = 7.6 (95% CI: 1.5–37.3), p = 0.006). Presence of vascular calcifications was also associated with an SBAC-L1 ⩽ 145 HU (OR = 8 (95% CI: 2.5–25.5), p = 0.0001).

Table 2.

Specific risk factors for an SBAC-L1 ⩽ 145 HU in univariate and multivariate analysis.

SBAC-L1 N = 70	>145 HU N = 35 (50%)	⩽145 HU N = 35 (50%)	Univariate		Multivariate
			p	OR (95% CI)	p	OR (95% CI)
Demographic/clinical data
Age (SD, years)	54.7 ± 15	70 ± 12.2	0.0001	1.1 (1.0–1.1)/year	0.02	1.1 (1–1.1)
Disease duration (years)	9 (5–11)	10 (8–14.5)	0.05	1.1 (1.0–1.2)	0.4	1.1 (0.9–1.2)
Sex (male)	3 (8.6)	5 (14.3)	0.5	0.6 (0.1–2.6)
Smoking	8 (22.9)	9 (25.7)	0.78	1.2 (0.4–3.5)
Obesity (BMI > 30 kg/m²)	5 (14.3)	4 (11.4)	0.72	0.8 (0.2–3.2)
Systemic scleroderma	9 (25.7)	7 (20)	0.57	0.7 (0.2–2.2)
Osteoporosis
⩾ 1 clinical risk factor	25 (71.4)	35 (100)	0.35	1.248 (0.8–2.0)
VF	1 (2.9)	2 (5.7)	0.6	2.1 (0.2–23.8)
Vitamin calcic supplementation (n = 40)	3 (8.6)	15 (42.9)	0.035	5.0 (1.12–22.30)	0.16	3.5 (0.6–20.3)
Spinal calcifications (binary)
Anterior	24 (68.6)	33 (94.3)	0.006	7.6 (1.5–37.3)	0.6	1.6 (0.2–11)
Posterior	12 (34.3)	12 (34.3)	1	1 (0.4–2.7)
Canal	12 (34.3)	10 (28.6)	0.607	0.77 (0.3–2.1)
Foraminal	4 (11.4)	4 (11.4)	1	1 (0.2–4.4)
Vascular calcifications (binary)	15 (42.9)	30 (85.7)	0.0001	8 (2.5–25.5)	0.4	1.8 (0.4–8)
Aortic valvular calcification	8 (22.9)	13 (37.1)	0.210	1.99 (0.7–5.8)
Mitral valvular calcifications	8 (22.9)	10 (28.6)	0.576	1.37 (0.5–4.1)

Qualitative data are described by the count (percentage), whereas quantitative data are described by the average ± SD (standard deviation) for age or by the median and interquartile index for other parameters. The duration of disease progression corresponded to the time between the date of diagnosis and the date of the performed imaging examinations. Age and disease duration are in years. Data in bold are statistically significant data (p<0.05).

95% CI: confidence interval of 95%; BMI: body mass index; OR: odds ratio; SBAC-L1: scanographic bone attenuation coefficient of the first lumbar vertebra; SD: standard deviation; VF: vertebral fracture; HU: Hounsfield units.

In multivariate analysis (Table 2), only age was significantly associated with a lower SBAC-L1, with an OR of 1.1 (95% CI: 1–1.1, p = 0.026). The presence of vascular calcifications or anterior spinal calcifications was not associated with an SBAC-L1 ⩽ 145 HU (p = 0.182 and 0.486, respectively).

Discussion

In this study, we showed that 64.3% of systemic sclerosis patients had aortic vascular calcifications; 30% and 25.7% had aortic or mitral valvular calcifications, respectively; and 84.3% had at least one spine calcification. There was no difference between the patients with diffuse or limited disease for ectopic calcifications (aortic vascular, aortic or mitral valvular or spinal). In multivariate analysis, only age was significantly associated with a lower SBAC-L1, with an OR of 1.1 (95% CI: 1–1.1, p = 0.026). The presence of vascular calcifications or anterior spinal calcifications was not associated with an SBAC-L1 ⩽ 145 HU. We did not find any relationship between corticosteroid intake and bone fragility. This finding is in agreement with the literature in systemic sclerosis,^8,43 probably due to a lower cumulative dose of corticosteroids in patients with systemic sclerosis (with short period of time and/or low doses, to avoid renal involvement).

Previously, we demonstrated that the measure of trabecular bone mass on CT (SBAC-L1 in HU) was highly reproducible in a population with systemic sclerosis. ²⁶ We showed that osteoporosis screening was not systematically performed (only 42.8% of patients had undergone DXA) in these patients with multiple risk factors for osteoporosis (at least one risk factor for osteoporosis in 85.7% of the patients). Fifty percent of the systemic sclerosis patients presented an SBAC-L1 under the fracture threshold, which suggested an underestimated risk of fracture using DXA (osteoporosis prevalence at 16.7%).

In a previous study, ²⁶ we showed that calcinosis and periarticular and vascular calcifications were significantly associated with an SBAC-L1 ⩽ 145 HU (OR = 6.30 (95% CI: 1.61–24.75), p = 0.004; OR = 3.22 (95% CI: 1.06–9.77), p = 0.03; and OR = 8 (95% CI: 2.5–25.5), p = 0.0001, respectively). Interestingly, in this study, we showed that an SBAC-L1 value under the fracture threshold was strongly associated with age but not with the presence of spinal, valvular or vascular calcifications in multivariate analysis. This could be due to a lack of power because of the low number of patients. Previously, we also showed that the presence of calcinosis, periarticular calcifications and acro-osteolysis on hand radiography was linked with low SBAC-L1, especially for patients under 63 years. ²⁶

Some studies describe a relation between calcinosis and periarticular calcifications, acro-osteolysis and osteoporosis on DXA in systemic sclerosis.^{10,19,21,24,44} Osteoporosis and vascular calcification have been considered independent processes related to age, although some recent studies have shown a close association between vascular and/or valvular calcifications and osteoporosis.^{36,45 –49} The RANKL/RANK/OPG system and the Wnt/β-catenin signalling pathway are the biological links between the bone and vascular systems and are regulated by plasma concentrations of osteoprotegerin (OPG) and sclerostin.^44,45

The threshold of 145 HU was chosen because it allowed the best compromise between sensitivity and specificity. ³³ In our population, with only three patients presenting VFs, we were unable to calculate the optimum SBAC-L1 threshold for VFs in terms of sensitivity and specificity. In a previous study, we showed that an SBAC-L1 ⩽ 145 HU was associated with a higher risk of a VF, similar to a T-score ⩽ –2.5 SD on DXA, with an OR approximately 2. ⁵⁰

We have previously shown that the SBAC-L1 could be easily used in patients with inflammatory rheumatic diseases, such as rheumatoid arthritis or ankylosing spondylitis. ⁴⁸ CT offers the opportunity for pertinent qualitative and quantitative screening for bone fragility. This method permits us to screen more patients with bone fragility than DXA. A low SBAC-L1 (under the fracture threshold) permitted the identification of more patients with VFs than osteoporosis defined on DXA (T-score under 2.5 SD).^26,33,48

The SBAC-L1 measure could be used in clinical practice. An ROI could be drawn on the software, and the trabecular bone density was automatically calculated. Furthermore, with this measure, we can exclusively explore trabecular bone without cortical bone or artefacts due to 2D projection, as observed with DXA, such as vascular calcification, spinal osteoarthritis, bone condensation or disc/vascular calcifications.^{33 –37} DXA remains the gold standard for osteoporosis screening, but it is not sufficiently performed in systemic sclerosis patients, including 40% of patients treated with corticosteroids. CT is usually performed in systemic sclerosis patients to explore systemic involvement of the disease. We can also perform an opportunistic screening of osteoporosis on CT to obtain information concerning the trabecular bone mineralization of the first lumbar vertebra in patients for whom CT was performed during the follow-up of their disease (according to good clinical practice), whereas DXA was unavailable. A limit of this opportunistic screening is the available segment for the vertebral morphology study: from C7 to L1 for thoracic CT and from C7 to S1 for TAP CT. We can also miss some VFs on lumbar vertebrae on thoracic CT.

The limitations of this study were the low number of patients and its retrospective nature. Furthermore, we can create a bias with the choice of patients with CT because they could have more active disease than patients without CT. Patients with only thoracic CT could not be explored for FVs on the lumbar segment and for vascular calcifications on the terminal aorta.

Inter- and intra-reader reproducibilities for ectopic calcifications were good to very good, but the confidence intervals were very large due to the low number of patients. We have to reproduce these findings in other studies with a larger population.

These results are interesting, but they must be confirmed in a larger population of patients with systemic sclerosis, and the pathophysiological mechanisms involved in both bone fragility and ectopic calcifications must be specified. It would be interesting, in a future study, to compare SBAC-L1 values between patients with systemic sclerosis and healthy donors, matched for age and gender. Molecular mechanisms associated with osteoblast activation and the repression of bone resorption (the RANKL/RANK/OPG system and the Wnt/β-catenin signalling pathway) can become the target of therapy for osteoporosis and calcification changes.

Conclusion

To conclude, we showed that the presence of ectopic calcifications was not associated with the SBAC-L1 under the fracture threshold in patients with systemic sclerosis.