We included 142 consecutive patients with sarcoidosis according to the criteria retained in the Consensus Conference ATS/ERS 14 , that is, combining clinical, biological and radiological presentation compatible with diagnosis and excluding other granulomatous diseases. Among 222 patients attending pneumonology consultation or day-hospital during the inclusion period, 74 were excluded because they did not meet inclusion criteria and 6 refused to participate in the study. All 142 remaining included patients presented documented histological lesions of granuloma without caseous necrosis, in at least one site of biopsy. Patients with other chronic progressive diseases, chronic respiratory or renal insufficiency stage IV and V whose origin was not related to sarcoidosis and those on diuretics able to interfere with calcium metabolism were excluded from the study. At inclusion, all patients underwent clinical examination evaluating the risk factors for osteoporosis and calcium intake. All patients had biochemical, radiological and BMD assessment.

This study complies with the Declaration of Helsinki and was approved by the Ethical Committee of France 10 (N° ID RCB 2011-A00202-39). All patients gave their informed consent prior to their inclusion in the study.

For each patient, a questionnaire assessed the following risk factors for osteoporosis: age, sex, menopausal status, tobacco and alcohol consumption, body mass index (BMI), personal or family history of fracture of low energy (defined as resulting from a fall from standing height or lower; skull, metacarpal and metatarsal fractures were excluded).

Anti-osteoporotic treatments, current and cumulated CS dose and vitamin D supplements in the six months preceding the study were recorded. Dietary calcium intake was evaluated by the Fardellone auto-questionnaire 15 .

The data related to sarcoidosis were also collected: disease duration, localization (graded from 0 to 6 according to the degree of severity of each organ involvement), current flare (at least one active localization or new localization in the three months preceding the study), number of relapses, stage of pulmonary involvement (stage 0: normal chest radiography; stage I: bilateral hilar lymphadenopathy without pulmonary infiltrates; stage II: bilateral hilar lymphadenopathy with pulmonary infiltrates; stage III: pulmonary infiltrates without bilateral hilar lymphadenopathy; stage IV: end-stage fibrosis, bullae, honeycombing and cavity), stage of dyspnoea from I to IV according to New York Heart Association (NYHA) classification.

The serum levels of calcium (corrected with albumin), phosphate, 25(OH)D (Diasorin radioimmunoassay, Stillwater, US), 1,25(OH)2D (IDS radioimmunoassay, Frankfurt, Germany), creatinine, parathyroid hormone (PTH) (immunometry LIA, Immulite 2000, Puteau, France), thyroid stimulating hormone (TSH), bone markers: bone alkaline phosphatases (BALP) (ELISA, reactif microrevue), C-terminal telopeptide of type I collagen (CTX) (ECLIA/Cobas-Roche, Switzerland), osteocalcin (immunometry TRACES, Kryptor), 24-hour urinary calcium, phosphate and creatinine, were gathered. Other parameters: erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), serum angiotensin-converting enzyme (ACE) level, blood count and serum albumin level were also gathered.

Radiographs of thoracic and lumbar spine (face and profile) were carried out. Assessment of vertebral fractures was independently performed by both a radiologist (NN) and a rheumatologist (NSK) according to the semi-quantitative method of Genant 16 and their respective results were blinded during assessment. Inter-rater reliability was good K = 0.78, 95% CI (0.63 to 0.93). Vertebral fracture was defined by reduced height loss above 20% of the mean, posterior or anterior wall.

BMD was measured by dual X-ray absorptiometry (DXA Lunar Prodigy, GE Healthcare) at the lumbar spine and the total hip. Osteopenia and osteoporosis were defined by a-1SD < T-score < -2.5 SD and T-score ≤ -2.5 SD, respectively. Low BMD was defined by a T-score < -1 SD (osteopenia or osteoporosis).

Data are summarized as the mean and standard deviation for continuous data and frequency for categorical data. A logistic regression model was used to identify factors associated with low BMD (defined as a T-Score lower than -1 SD) and risk of fracture. 25(OH)D was grouped in three categories as retained by the Institute of Medecine 17 : ≤10 ng/ml (deficiency), 10 to 20 ng/ml (insufficiency) and ≥20 ng/ml (desirable). All factors with P <0.20 at univariate analysis were included in a multiple logistic regression model with backward selection. Age, gender and menopause were grouped together to avoid a colinearity problem in multivariate analysis. Serum 25(OH)D and calcium levels according to different criteria were compared with the Mann-Whitney U-test, with a Bonferroni correction for multiple tests. The associations between the continuous factors were determined with Spearman’s correlation coefficients. All tests were two-sided at a 0.05 significance level. Analyses were carried out using R statistical software version 2.14.1.

One hundred forty two patients, 80 women (51 menopaused) and 62 men, were included. Eighty-five patients were Caucasians, 54 Caribbeans and 3 Indians. Mean age was 51.6 ± 11.6 years and BMI 27.5 ± 5.4. Mean disease duration was 9.5 ± 7.1 years and 104 patients had presented more than one relapse; 21/137 patients experienced a low energy fracture before inclusion. Eighty-eight patients were receiving CS treatment at the time of the study (mean dose 12.4 ± 11.8 mg/d of prednisone equivalent), 28 patients had never received CS, 45 interrupted CS treatment for at least six months. The mean cumulated CS dose was 27.6 ± 19.9 g of prednisone equivalent. Thirty-one patients had received vitamin D supplements in the six months preceding the study (mean dose 181,161 ± 137,430 U) and 15 were on supplementation at inclusion. Forty-six patients had received a specific bone treatment before the study (bisphosphonates in the majority) on average for 25.3 ± 26.7 months; 24 were still treated during the study. The mean daily dietary calcium was 717.2 ± 360 mg (Table 1).

ACE, angiotensin-converting enzyme; BMI, Body mass index; BP, bisphosphonates; CRP, C-reactive protein; CS, Corticosteroids; ESR, erythrocyte sedimentation rate; NYHA, New York Heart Association; PTH, parathyroid hormone.

Biochemical parameters are shown in Table 1. Serum calcium was within the normal range with no significant difference between summer (June to August) and winter (December to February), respectively, with a mean level of 2.36 ± 0.06 mmol/l vs 2.32 ± 0.1 mmol/l ( P = 0.16). Only one patient had high serum calcium (>2.6 mmol/L), (without hypercalciuria) related to a systemic form with documented bone sarcoidosis lesions. Nine out of 94 patients presented hypercalciuria (24-hour urinary calcium >0.1 mmol/kg).

Serum 25(OH)D levels were low (Figure 1) with a mean of 14.5 ± 7.61 ng/ml but with normal mean PTH serum level. The 31 patients who had received vitamin D supplements had significantly higher 25(OH)D but not higher serum calcium levels vs. those not supplemented and no significant change in 1,25(OH)2D serum level (Figure 2).

CS-treated or untreated patients did not differ in 25(OH)D and 1,25(OH)2D serum levels nor in CRP or ESR (trend toward a lower ESR on CS).

There was no association between bone markers and BMD value or fracture. However, as expected, CS-treated patients had significantly lower bone marker serum levels: (CTX : 372.1 pg/ml with CS vs. 480.6 pg/ml without CS, P = 0.01; BALP : 30.0 ± 13.5 UI/L vs 33.5 ± 10.8 UI/L, P <0.05; osteocalcin: 15.5 ± 10.1 ng/ml vs 25.2 ± 13.2 ng/ml, P <0.001).

Vitamin D supplementation was positively correlated with serum 25(OH)D (r = 0.38, P <0.001) but not with serum calcium level.

Serum levels of 25(OH)D show significant but weak inverse correlation with disease flares (r = -0.18, P <0.05), the severity of the pulmonary involvement (r = -0.18, P <0.05), ACE (r = -0.19, P <0.05) and ESR (r = -0.19, P <0.05). 1,25(OH)2D serum levels are positively correlated with serum 25(OH)D (r = 0.35, P <0.001) and with no other parameter. In particular, we did not find correlation between 1,25(OH)2D serum level and chronicity of the disease (r = -0.08, P = 0.37) that was found by D Kavathia et al. 18 or the number of relapses.

The mean BMD was normal: Tscore: -0.5 SD at the lumbar spine and -0.09 SD at the total hip. However, 34.8% presented low BMD at the lumbar spine (37 patients had osteopenia and 11 patients had osteoporosis, respectively) and 24% had osteopenia or osteoporosis at the hip (24 patients and 9 patients, respectively). On the whole, 40.15% (55/137) were osteopenic at one skeletal site at least, and 14.6% (20/137) were osteoporotic.

A total of 13.6% of the patients (19/139) had at least one vertebral fracture and 10 patients presented with two or more vertebral fractures. Overall, 23.5% of patients (32/136) had at least one vertebral or peripheral fracture.

Table 2 shows the odds ratio of low BMD and fractures at univariate analysis. Low BMD (T-score < -1 SD) is associated with age, menopause, prevalent fracture, low dietary calcium intake, cumulative CS dose, long disease duration, advanced-stage dyspnoea (III or IV), lymphopenia, high ESR and low creatinine clearance. The patients with serum levels of 25(OH)D between 10 and 20 ng/ml have the lowest odds of low BMD, whereas the odds increase when this threshold is exceeded. Fractures were significantly associated with age, low dietary calcium intake, cumulative and current CS dose, advanced-stage dyspnoea (III or IV), low creatinine clearance and low BMD. Of note, 25(OH)D levels exceeding 20 ng/ml are associated with significantly higher odds of fracture (Table 2). BMI, ethnicity, type of sarcoidosis involvement, CS-free period duration, smoking, 1,25(OH)2D serum level, CRP, bone remodelling markers showed no association with BMD or fracture.

BMD, bone mineral density; BP, bisphosphonates; CS, Corticosteroids; ESR, erythrocyte sedimentation rate; PTH, parathyroid hormone; NYHA, New York Heart Association.

We then assessed factors associated with low BMD using multivariate analysis (Table 3). Age, prevalent fracture, female gender, menopause, low dietary calcium, lymphopenia and vitamin D supplementation are associated with higher odds of low BMD. Again, the patients with 25(OH)D serum levels between 10 and 20 ng/ml are at lower risk of low BMD vs. those with levels <10 ng/ml (reference class), while levels exceeding 20 ng/ml are associated with a higher risk of fracture. Low dietary calcium and high CS doses are also associated with a higher risk of fracture.

BMD, Bone mineral density; BP, Bisphosphonate; CS, Corticosteroids. Empty cells correspond to variables not included in the multivariate analysis because they were not significant at univariate analysis or to variables included in the multivariate analysis but not retained in the final model because they were no longer significant (for example, age).

The relationship between 25(OH)D serum level and risk of low BMD in our population, using a generalized additive model followed a U-shaped curve: patients with serum 25(OH)D levels between 10 and 20 ng/ml are at lower risk of low BMD. The relationship between 25(OH)D and fracture is more linear, but with a steeper slope for 25(OH)D values above 20 ng/ml (Figure 3A, B).

We then evaluated whether history of vitamin D supplementation or BP treatment modified the relationship between 25(OH)D serum level, low BMD and fracture. In our sample, vitamin D-supplemented or BP-treated patients had higher prevalence of bone fragility fracture (P <0.01) and lower BMD at both the lumbar spine (P <0.01) and hip (P <0.05) (Table 4).

BMD, bone mineral density; BP, bisphosphonates. The results are reported for the whole sample and for those patients that had not received BP treatment or vitamin D supplementation.

The relationship between 25(OH)D serum level and BMD is unmodified if those patients are excluded from the analysis (interaction between 25(OH)D serum level and vitamin D supplements or BP treatment, P = 0.71), (Figure 3C). Conversely, as far as fracture is concerned, there is a significant interaction between 25(OH)D and vitamin D and BP treatment (P <0.05), explained by the low prevalence of fracture in patients with serum levels of 25(OH)D <10 ng/ml if vitamin D and BP-treated patients are excluded (Table 4). Nevertheless, even after exclusion of these patients, the relationship between 25(OH)D levels and fracture remains linear (Figure 3D).