“Trabecular Bone Score” (TBS): An indirect measure of bone micro-architecture in postmenopausal patients with primary hyperparathyroidism

Bone 53 (2013) 154–159

Contents lists available at SciVerse ScienceDirect

Bone journal homepage: www.elsevier.com/locate/bone

Original Full Length Article

“Trabecular Bone Score” (TBS): An indirect measure of bone micro-architecture in postmenopausal patients with primary hyperparathyroidism☆ Elisabetta Romagnoli a,⁎, Cristiana Cipriani a, Italo Nofroni b, Claudia Castro a, Maurizio Angelozzi a, Addolorata Scarpiello a, Jessica Pepe a, Daniele Diacinti c, Sara Piemonte a, Vincenzo Carnevale d, Salvatore Minisola a a

Department of Internal Medicine and Medical Disciplines, University of Rome “Sapienza”, Rome, Italy Department of Public Health and Infective Diseases, University of Rome “Sapienza”, Rome, Italy c Department of Radiology, University of Rome “Sapienza”, Rome, Italy d Unit of Internal Medicine, “Casa Sollievo della Sofferenza”, IRCCS, San Giovanni Rotondo, Foggia, Italy b

a r t i c l e

i n f o

Article history: Received 8 August 2012 Revised 28 November 2012 Accepted 29 November 2012 Available online 7 December 2012 Edited by: Rene Rizzoli Keywords: Bone mineral density Primary hyperparathyroidism “Trabecular Bone Score” Vertebral fractures

a b s t r a c t Background: Patients with primary hyperparathyroidism (PHPT) generally show reduced bone mineral density (BMD) at cortical sites with relatively preserved trabecular bone. However, the increased fracture risk at all skeletal sites suggests that areal BMD probably is not effective in capturing all the determinants of bone strength. “Trabecular Bone Score” (TBS) has been recently proposed as an indirect measure of bone micro-architecture. Our study was aimed to investigate TBS in patients with PHPT. Methods: Seventy-three Caucasian postmenopausal women with PHPT and 74 age-matched healthy women (C) were studied. In all participants BMD at lumbar spine (LS) and at femoral sites (Neck-FN and total hip-TH) was measured by DXA and, in 67 patients and 34 C, also at the distal 1/3 of the radius (R). TBS was measured in the region of LS-BMD. Spine X ray was assessed in all patients. Results: Mean TBS values were significantly reduced in PHPT (1.19 ± 0.10) compared to C (1.24 ± 0.09, p b 0.01). Patients and controls did not differ for age, years since menopause (YSM), BMI, 25(OH)D serum levels, creatinine clearance, LS-BMD and FN-BMD. On the contrary, mean BMD values at both TH and R were significantly lower in PHPT patients compared to controls (p b 0.01 and p b 0.0001, respectively). In PHPT with vertebral fractures (VF +, n = 29) TBS was significantly lower than in those without fracture (VF−, n = 44)(1.14 ± 0.10 vs. 1.22 ± 0.10, respectively; p b 0.01), whose TBS values did not differ from C. Mean TBS values in patients with (n = 18) and without (n = 55) non-vertebral fractures did not significantly differ (1.16 ± 0.09 vs. 1.20 ± 0.11). The presence of vertebral fractures was independently associated with the reduction of TBS (OR = 0.003, 95% CI = 0–0.534, p = 0.028) and with YSM (OR = 1.076, 95% CI = 1.017–1.139, p = 0.011), but not with age, the reduction of LS-BMD and the increase of BMI. The combination of YSM > 10 years plus TBS b 1.2 was associated with a significant risk of VF (OR = 11.73, 95% CI 2.43–66.55, p b 0.001). A TBS value b1.2 showed a better performance in individuating VF (sensibility 79.3%, specificity 61.4%, positive predictive value 57.5%, and negative predictive value 81.8%) in respect to YSM > 10 years. Conclusions: TBS seems to indirectly reflect an alteration of bone micro-architecture in postmenopausal women with PHPT. © 2012 Elsevier Inc. All rights reserved.

Introduction

☆ Disclosure: All authors state that they have no conflicts of interest. ⁎ Corresponding author at: Department of Internal Medicine and Medical Disciplines, University of Rome “Sapienza”, Viale del Policlinico 155, 00161 Rome, Italy. Fax: +39 6 45549053. E-mail addresses: [email protected] (E. Romagnoli), [email protected] (C. Cipriani), [email protected] (I. Nofroni), [email protected] (C. Castro), [email protected] (M. Angelozzi), [email protected] (A. Scarpiello), [email protected] (J. Pepe), [email protected] (D. Diacinti), [email protected] (S. Piemonte), [email protected] (V. Carnevale), [email protected] (S. Minisola). 8756-3282/$ – see front matter © 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bone.2012.11.041

Primary hyperparathyroidism (PHPT) is a common endocrine disease, more frequently affecting postmenopausal women [1,2], whose clinical presentation has changed over time. In particular, whereas osteitis fibrosa cystica is seen quite rarely [3], bone involvement is nowadays characterized by various degrees of bone loss. A lower bone mineral density (BMD) at cortical sites can be demonstrated by means of dual-energy X ray absorptiometry (DXA), while trabecular bone mass is relatively preserved [4,5]. Accordingly, bone biopsy studies showed cortical thinning and increased cortical porosity but maintenance of cancellous bone volume [6]. The preservation

E. Romagnoli et al. / Bone 53 (2013) 154–159

of trabecular tissue is also indicated by the 3-dimensional analysis using μCT [7]. These findings collectively suggest that the effects of endogenous parathyroid hormone [PTH] excess on trabecular and cortical bone may be different [8]. Based on the aforementioned densitometric and histomorphometric data, fracture risk would be expected to be increased at skeletal sites mainly composed by cortical tissue in respect to those rich in trabecular bone. However, most of the evidence to date suggests that PHPT is associated with increased fracture risk at both trabecular and cortical skeletal sites [9–12]. In particular, data from our and other groups demonstrated that vertebral fracture risk is increased in postmenopausal women with PHPT [13,14], even when lumbar spine BMD is preserved. It has been hypothesized that such results may at least partly depend on surveillance bias, since these patients undergo spine X rays more likely than the general population. On the other hand, the thinning of the cortical envelope of vertebral bodies, as well as the high bone turnover rate [15], trabecular number, thickness, connectivity, and bone geometry may be important determinants of bone strength other than BMD. The relative contribution of these factors may be explored by non-invasive imaging techniques such as e.g. high-resolution peripheral quantitative computed tomography (HR-pQCT) and high-resolution magnetic resonance (HR-MRI) [16]. The use of these techniques in clinical practice is limited by multiple factors, i.e. costs, time spent to perform the investigation and, for HR-pQCT, the need of a dedicated scanner. The HR-pQCT also allows to in vivo evaluate several parameters usually investigated through invasive techniques, as well as bone geometry and volumetric BMD [17]. HR-pQCT showed an alteration in geometry, volumetric density, and micro-architecture of both trabecular and cortical components of the radius, but not of the tibia, in postmenopausal women with PHPT [18,19]. However, this technique suffers from motion artifacts, which may significantly impair the image quality and parameter calculations; furthermore, it allows only the investigation of peripheral sites. The “Trabecular Bone Score” (TBS) is a recently introduced tool which has been reported to indirectly reflect bone micro-architecture [20]. Indeed, it is a texture parameter seemingly recording pixel gray-level variations in DXA images [21]. Previous studies demonstrated that TBS positively correlates with 3D bone micro-architecture parameters, such as connectivity density and trabecular number, and negatively with trabecular separation [21–23]. In clinical practice, the software for TBS computation can be installed on DXA machines, and TBS is automatically calculated consecutively on BMD measurement. Thus, TBS has been claimed to be a reproducible and easy to handle quantitative value, whose higher scores reflect stronger and more fracture-resistant micro-architecture, whereas lower scores indicate poor bone quality and greater fracture susceptibility. Cross-sectional studies indicate that TBS in addition to BMD measurement might be employed for fracture risk assessment [24]; TBS is able to discriminate osteoporotic patients with fracture from those without fracture, and to predict major osteoporotic fractures. In fact, in postmenopausal women with a previous osteoporotic fracture TBS was lower than in age- and BMD-matched non-fractured women [25–27] and, in a retrospective analysis of the Manitoba Study, TBS predicted osteoporotic fractures independently of bone density [28]. Finally, other studies documented the added value of TBS in evaluating fracture risk in patients with secondary causes of osteoporosis [29,30]. To our knowledge, no data are available so far exploring the clinical utility of TBS to assess skeletal involvement and fracture risk in patients with PHPT. This study was aimed to investigate this issue in postmenopausal women with PHPT, whose estrogen deficiency is expected to preferentially affect cancellous bone, with a consequent increased risk of vertebral fractures. Subjects and methods Seventy-three consecutive postmenopausal Caucasian patients with PHPT, coming to our Mineral Metabolism Center from January 2010 to

155

December 2011, were enrolled in the study. The diagnosis of PHPT was made according to the conventional clinical and laboratory criteria, including a history of at least 1 year of prolonged hypercalcemia without evidence of a non-parathyroid etiology and unsuppressed serum levels of immunoreactive PTH. Familial hypocalciuric hypercalcemia was excluded because in all cases the ratio of calcium to creatinine clearance was b0.01. Clinical diagnosis was confirmed in 48 patients by the removal of a histologically proven parathyroid adenoma. No patients took medications (including estrogen and bisphosphonates) known to influence mineral metabolism. Seventy-four healthy postmenopausal Caucasian women, matched for age, years since menopause (YSM) and body mass index (BMI) were concomitantly studied as a control group (C). They were randomly selected from a group of ambulatory postmenopausal women referred in the same period to our Center by general practitioners for a menopause-screening program. In each of them, medical history and physical examination excluded disorders of bone and mineral metabolism, and they did not take medications affecting bone (including estrogen and bisphosphonates). In all the participants a 24-h urine collection along with a blood sample were collected in order to measure the main parameters of calcium metabolism, as well as PTH and 25-hydroxyvitamin D (25OHD) serum levels, according to previously described methods [31]. All patients had standardized radiographs in antero-posterior and left lateral projections of the thoracic and lumbar spine, centered at T8 and L3 respectively, at a film-focus distance of 115 cm. The radiographs were examined first for quality and then for fractures by an experienced skeletal radiologist. Vertebral fractures were defined using the Genant's semiquantitative method (combination of morphometric and visual assessment), which is commonly used for diagnosis of vertebral fractures [32]. According to this method, Grade 1 (mild) fracture is a reduction in vertebral height of 25%, Grade 2 (moderate) a reduction of 26–40% and Grade 3 (severe) a reduction of >40%. Non-vertebral fractures were recorded at history. X ray of the spine was not systematically performed in control subjects due to ethical reasons; only 15 of them had previously undergone X ray examination for back pain. BMD was measured by DXA (QDR-4500; Hologic Inc., Waltham, MA) at the lumbar spine in posterior-anterior projection (L1–L4) (LS-BMD) and at two femoral sites [femoral neck (FN-BMD) and total femur (TH-BMD)] in all patients and controls. BMD of the nondominant forearm at the distal 1/3 of the radius (R) was also assessed in 67 out of 73 PHPT patients and 34 of 74 controls. Fractured vertebrae were excluded from BMD measurement. The coefficients of variations were 1.0% at lumbar spine, 1.5% at femoral neck, 1.7% at total hip and 1.3% at the proximal third of the radius. In all patients and controls TBS was evaluated in the same regions as those used for LS-BMD (L1-L4) using TBS iNsight® (Version 1.8, Med-Imaps, Pessac, France). TBS was calculated as the mean value of the individual measurements for vertebrae L1–L4. Vertebrae excluded for BMD assessment were also excluded for TBS evaluation at lumbar spine. The coefficient of variation for TBS was 1.8%, and it did not vary among the measured vertebrae. The study was approved by the local Ethics Committee. All subjects gave informed consent. Statistical analysis Descriptive statistics are expressed as the mean ± SD. After a normality test, comparisons between groups were performed by t-test. A forward stepwise logistic regression analysis, including age, YSM, BMI, LS-BMD, and TBS as covariates, was used to identify potential predictors of vertebral fractures. The Receiver Operating Characteristics (ROC) curve analysis was performed to test the ability of TBS, YSM and BMD at various skeletal sites to predict vertebral fractures. The best cut-offs of YSM and TBS in predicting vertebral fractures were defined based on ROC curve analysis. Utilizing these cut-offs, sensitivity (SN), specificity (SP), positive predictive value (PPV), and negative predictive value (NPV) for YSM and TBS were calculated.

156

E. Romagnoli et al. / Bone 53 (2013) 154–159

The odd ratios (OR) for detecting vertebral fractures of the combination of YSM plus TBS were also calculated. Statistical significance was set at p b 0.05. Statistical analysis was performed by SPSS version 12.0 statistical package (SPSS Inc., Chicago, IL).

Results Main demographic, anthropometric, biochemical and densitometric data of patients with PHPT and controls are reported in Table 1. Mean values of age, YSM, height, weight, BMI, 25(OH)D serum levels, creatinine clearance, LS-BMD and FN-BMD were not significantly different between patients and controls, while mean BMD values at both total hip and radius were significantly lower in patients compared to controls (p b 0.01 and p b 0.0001, respectively). As expected, mean values of serum calcium, parathyroid hormone levels, and 24 h urinary calcium were significantly higher in PHPT patients compared to C. Notably, mean values of TBS in PHPT patients were significantly lower than those found in controls (p b 0.01) (Table 1). We then subdivided PHPT patients into two subgroups according to the absence (VF−, n = 44) or the presence of vertebral fractures (VF+, n = 29). Twenty-four patients had Grade 1, 4 patients had at least one Grade 2, and 1 patient had a Grade 3 deformity; 17 patients had only one VF whereas 12 patients had more than 1 VF. Twenty-three patients were asymptomatic, while six patients had clinical vertebral fractures. The prevalence of vertebral fractures was not assessed in controls since X ray was not systematically performed in this group (one vertebral fracture was detected in 4 out of 15 subjects in which an X ray was available). Mean values of TBS were significantly lower in VF + patients in respect to those found in C (1.14 ± 0.10 vs. 1.24 ± 0.09, p b 0.001) whereas mean values of TBS of VF − patients did not differ from those of C (1.22 ± 0.10 vs. 1.24 ± 0.09, p = NS). Mean values of TBS in VF + patients were significantly lower than those found in VF − patients (p b 0.01). TBS values did not differ between patients with one VF (n = 17) and those with more than 1 VF (n = 12) (1.14 ±

Table 1 Mean values ± 1 SD of demographic, anthropometric, biochemical and densitometric parameters in patients with PHPT and controls. Parameter

0.12 vs. 1.15 ± 0.09, p = NS). Furthermore, patients with vertebral fractures were also significantly older (p b 0.01), had a significantly longer YSM period (p b 0.001), a higher BMI (p b 0.05) and a reduced BMD at radius (p b 0.05) in respect to VF − patients. The two subgroups did not differ for the mean values of LS-BMD, FN-BMD and TH-BMD (Table 2). The mean TBS values in patients with (n = 18) and without (n = 55) non-vertebral fractures did not significantly differ (1.16 ± 0.09 vs. 1.20 ± 0.11). The Logistic Regression analysis showed that the presence of vertebral fractures in PHPT patients was independently associated with both the reduction of TBS (OR = 0.003, 95% CI = 0–0.534, p = 0.028) and the increase of YSM (OR = 1.076, 95% CI = 1.017–1.139, p = 0.011), but not with advancing age, the increase of BMI and the reduction of LS-BMD. The ROC curve analysis showed that both TBS and YSM did significantly predict vertebral fractures (AUC 0.716, 95% CI = 0.59–0.841, p = 0.002 and AUC 0.717, 95% CI 0.595–0.840, p = 0.002, respectively) (Fig. 1). The cut-offs with the best compromise between sensibility (SN) and specificity (SP) by ROC curve analysis were set at 1.2 for TBS and at 10 years for YSM. We employed these cut-offs in order to calculate the OR for detecting vertebral fractures in 3 groups of patients, classified according to the combination of YSM plus TBS (Table 3). The group with YSM>10 years plus TBS b 1.2 had an OR of 11.730 (pb 0.001) compared to the group with YSM b 10 years plus TBS> 1.2 (reference group). Table 4 shows the sensitivity (SN), specificity (SP), positive predictive value (PPV) and negative predictive value (NPV) of YSM and TBS in identifying PHPT patients with vertebral fractures. The TBS showed the better performance; indeed, a cut-off b 1.2 had a specificity of 61.4% and a NPV of 81.8% for excluding vertebral fractures, and a sensitivity of 79.3% and a PPV of 57.5% for detecting vertebral fractures. The discriminative value for vertebral fractures of TBS and BMD of various skeletal sites was also assessed through the area under the ROC curve. Only the area under the curve (AUC) of TBS was significant (0.731, 95% CI 0.604–0.857, p = 0.001). The remaining AUCs were 0.622 (95% CI 0.484–0.760, p = NS) for LS-BMD, 0.453 (95% CI 0.310–0.596, p = NS) for FN-BMD, 0.497 (95% CI 0.350–0.643, p = NS) for TH-BMD and 0.637 (95% CI 0.504-0.770, p = NS) for R-BMD, respectively (Fig. 2). The ROC curve showed an inflexion point for a sensitivity value of 80% and a specificity of 60% for TBS = 1.2.

PHPT patients (n = 73) Controls subjects (n = 74) p

Age (years) 63.6 ± 9.2 YSM (years) 14.5 ± 9.9 Height (cm) 159 ± 7.3 Weight (kg) 65.2 ± 12 BMI 25.8 ± 4.9 sCa (mg/dl) 10.8 ± 0.40 PTH (pg/ml) 86.4 ± 35.2 25(OH)D (ng/ml) 22.3 ± 9.3 uCa (mg/24 h) 201 ± 100.4 ClCr (ml/min) 83.3 ± 24 TBS 1.19 ± 0.10 LS-BMD (mg/cm2) 828.6 ± 139.6 T-score −1.98 ± 1.25 FN-BMD (mg/cm2) 639.8 ± 98.3 T-score −1.87 ± 0.88 TH-BMD (mg/cm2) 760.8 ± 112.5 T-score −1.44 ± 1.0 R-BMD (mg/cm2)* 574.1 ± 72.6 T-score −1.9 ± 1.23

61.3 ± 5.6 12 ± 7.8) 158 ± 6.5 63.9 ± 10.1 25.4 ± 4.0 9.51 ± 0.50 27.9 ± 9.5 23 ± 8.4 154.3 ± 78.9 90.1 ± 19.0 1.24 ± 0.09 866.1 ± 112.6 −1.62 ± 1.02 668.4 ± 91.3 −1.62 ± 0.82 808.5 ± 97.0 −1.09 ± 0.78 638.6 ± 52.9 −0.92 ± 0.88

NS NS NS NS NS b0.001 b0.001 NS b0.01 NS b0.01 NS NS NS NS b0.01 b0.05 b0.0001 b0.0001

YSM = years since menopause; BMI = body mass index; sCa = serum calcium levels; PTH = serum parathyroid hormone levels; 25(OH)D = serum 25-hydroxy-vitamin D levels; uCa = 24 h urinary calcium excretion; ClCr = creatinine clearance; TBS = “Trabecular Bone Score”; LS-BMD = bone mineral density at lumbar spine; FN-BMD = bone mineral density at femoral neck; TH-BMD = bone mineral density at total hip; R-BMD = bone mineral density at the distal 1/3 of the radius (*measured only in 67 patients and 34 controls, respectively).

Table 2 Mean values ± 1 SD of demographic, anthropometric and densitometric data in patients with PHPT subdivided according to the presence (VF+) or absence (VF−) of vertebral fractures. Parameter

VF + (n = 29)

VF− (n = 44)

P

Age (years) YSM (years) Height (cm) Weight (kg) BMI TBS LS-BMD (mg/cm2) T-score FN-BMD (mg/cm2) T-score TH-BMD (mg/cm2) T-score R-BMD (mg/cm2)* T-score

67.6 ± 8.2 19.2 ± 10.3 157 ± 5.8 67.8 ± 14.6 27.4 ± 6.2 1.14 ± 0.10 795.0 ± 134.0 −2.29 ± 1.21 642.3 ± 112.4 −1.85 ± 1.01 762.0 ± 126.5 −1.36 ± 1.2 550.7 ± 69.1 −2.34 ± 1.23

61.0 ± 8.9 11.5 ± 8.3 160 ± 7.9 63.6 ± 9.7 24.8 ± 3.6 1.22 ± 0.10 850.7 ± 140.3 −1.78 ± 1.25 638.1 ± 89.2 −1.88 ± 0.80 760.0 ± 10.38 −1.49 ± 0.85 590.0 ± 71.4 −1.73 ± 1.18

b0.01 b0.001 NS NS b0.05 b0.01 NS NS NS NS NS NS b0.05 b0.05

YSM = years since menopause; BMI = body mass index; TBS = “Trabecular Bone Score”; LS-BMD = bone mineral density at lumbar spine; FN-BMD = bone mineral density at femoral neck; TH-BMD = bone mineral density at total hip; R-BMD = bone mineral density at the distal 1/3 of the radius (*measured only in 27 patients with vertebral fractures and 40 patients without vertebral fractures, respectively).

E. Romagnoli et al. / Bone 53 (2013) 154–159

157

Table 4 Sensitivity (SN), specificity (SP), positive predictive value (PPV) and negative predictive value (NPV) for detecting vertebral fracture of YSM (cut-off 10 years) and TBS (cut-off 1.2).

YSM ≥ 10 95% CI TBS ≤1.2 95% CI

SN (%)

SP (%)

PPV (%)

NPV (%)

75.9 57.9-87.8 79.3 61.6–90.2

50 35.8–64.2 61.4 46.6–74.3

50 35.8–64.2 57.5 42.2–71.5

75.9 57.9–87.8 81.8 65.6–91.4

YSM = years since menopause; TBS = “Trabecular Bone Score”. The cut-offs for YSM and TBS were defined on the basis of the ROC curve analysis.

Fig. 1. Receiving operator characteristic curves of lumbar spine “Trabecular Bone Score” (TBS), and years since menopause (YSM) in vertebral fracture prediction, in the whole population of postmenopausal patients with primary hyperparathyroidism.

Discussion This is the first study investigating the clinical value of TBS to assess skeletal fragility in postmenopausal patients with PHPT. Our results show that mean TBS values are reduced in PHPT compared to a group of healthy women matched for age, YSM and anthropometric variables, though patients and controls did not differ for lumbar spine BMD. As expected, the mean BMD values measured at the cortical rich sites, such as the femur and the proximal third of the radius, were significantly lower in PHPT patients compared to controls. Taken together, these findings confirm the previously observed relative preservation of trabecular bone, as assessed by DXA, in patients with PHPT [3–5]. However, concerns have been raised on the ability of DXA to completely capture other important determinants of bone strength, apart from BMD, that could remarkably affect fracture risk. In this respect, our results deserve particular interest, since TBS has been proposed as an indirect index of bone micro-architecture, simple and easy to perform in clinical practice [20–23]. The reduction of mean TBS values in patients with PHPT demonstrates that bone quality is altered also at a skeletal site mainly composed by trabecular tissue, such as lumbar spine. This finding could partly account for the clinical observation of the increased vertebral fracture risk in these patients [9–14]. Other novel techniques have been recently proposed to provide “in vivo” access to parameters of bone quality [16,17]. Among them, HR-pQCT showed an alteration of geometry, volumetric density, and micro-architecture of both trabecular and cortical bone of the radius in postmenopausal women with PHPT [18,19]. However, several factors (as for instance costs, time spent to perform the investigation, the need for a dedicated scanner) and the possibility to explore only peripheral sites still limit a wide application of this technique in clinical practice. On the other hand, the assessment of TBS in

patients with PHPT could add some important information, even though indirectly estimated, on bone micro-architecture, as it has recently demonstrated for patients with other secondary causes of osteoporosis [29,30]. Moreover, the easy application of TBS software on DXA images could provide, without further assessment, two concomitant quantitative indexes of bone status for clinical practice purpose. The predictive value of TBS in detecting vertebral fractures was also assessed. The mean TBS values in PHPT patients without vertebral fractures did not differ from those of controls, which suggests that TBS could partially reflect texture integrity. Instead, the mean values of TBS were significantly lower in PHPT patients with vertebral fractures compared to those without VF, though their mean BMD values did not significantly differ at any site, but the radius. This last finding is particular relevant, because the antero-posterior DXA projection at the spine captures the vertebral shell in addition to trabecular bone, so that TBS measurement could also partially reflect cortical bone. This is in line with the reduced radial BMD in fractured PHPT patients. Patients with vertebral fractures were also older, had a

Table 3 Odd ratios (OR) for detecting vertebral fractures in PHPT patients by combining YSM plus TBS. The group with YSM b 10 years plus TBS > 1.2 was the reference group.

YSM b 10 years plus TBS b 1.2 YSM > 10 years plus TBS > 1.2 YSM > 10 years plus TBS b 1.2

OR

95% CI

P

1.62 3.5 11.73

0.25–11.26 0.46–26.62 2.43–66.55

NS NS b0.001

YSM = years since menopause; TBS = “Trabecular Bone Score”. The cut-offs for YSM and TBS were defined on the basis of the ROC curve analysis.

Fig. 2. Receiving operator characteristic curves of lumbar spine “Trabecular Bone Score” (TBS), and bone mineral density at lumbar spine (LS-BMD), at femoral neck (FN-BMD), at total hip (TH-BMD) and at the distal 1/3 of the radius (R-BMD) in vertebral fracture prediction, in the whole population of postmenopausal patients with primary hyperparathyroidism.

158

E. Romagnoli et al. / Bone 53 (2013) 154–159

significantly longer menopausal duration and a higher BMI than VF − patients. Our analysis showed that both YSM and TBS were significant predictors of vertebral fractures, after adjustment for potential confounders such as age, BMI and lumbar spine BMD. Both YSM and TBS had enough accuracy in predicting vertebral fractures when tested by ROC curve analysis. On the other hand, when different cut-offs of YSM and TBS were used to exclude vertebral fracture, the latter showed the best specificity (61.4%) and a NPV of 81.8%. Having YSM > 10 years plus a TBS value b1.2 also represented a strong combination for predicting vertebral fractures (OR 11.73, p b 0.001), whereas TBS is not predictive for patients with shorter menopausal duration. Finally, the ROC curve analysis also demonstrated that TBS was a better predictor of vertebral fracture in respect to BMD measured at different skeletal sites. Again, the cut-off showing the best compromise between sensitivity and specificity was identified with the same TBS value of 1.2. Such a cut-off appears to be slightly different from those reported by other authors in different clinical conditions [27,29]. This finding could partly rely to the different skeletal texture involvement in different diseases, so that the corresponding thresholds of TBS values are not directly comparable. Collectively taken, our results suggest that TBS may be a promising and simple clinical tool for detecting PHPT patients with higher risk for vertebral fracture, independently of BMD. In this respect, our findings deserve interest since the majority of patients we studied had asymptomatic vertebral fractures. This also implies that TBS seems to capture some aspects of bone fragility in these patients that probably are not accurately measured by areal BMD assessed by DXA. Actually, a low TBS value has been found to be associated with poor bone micro-architecture, low connectivity, high trabecular spacing, and a reduced number of trabeculae. Conversely, a high TBS value has been associated with good bone micro-architecture, high connectivity, low trabecular spacing and an augmented number of trabeculae [21–23]. These findings are in keeping with the results obtained by utilizing phalangeal ultrasound, which in turn indicates that some parameters, also reflecting bone mineralization but poorly correlated with BMD of the lumbar spine, may be altered in PHPT patients [33,34]. On the other hand, the lack of significant difference of TBS values between PHPT patients with and without non-vertebral fractures, seems to suggest that the technique could be regarded as site-specific tool. Therefore, the assessment of TBS in clinical practice could help to investigate bone strength in patients with PTH hormone excess, and also to identify patients deserving a more extensive evaluation of skeletal fragility. Our results also outline the importance of menopause duration that is of estrogen lack, rather than chronological age, in determining fracture risk in these patients. In other words, the longer is menopausal duration, the longer is the lack of the protective action of estrogen from the effects of sustained PTH excess on the skeleton. Previous studies showed that only age, or female gender, or BMD were significant independent predictors of fracture risk in PHPT patients [11,14,35]. A number of factors could account for these divergent results, such as differences in study-design, number of patients enrolled, co-morbidities, duration and severity of the disease, levels of bone turnover rate, and differences in vitamin D status [31,36]. Several studies demonstrated the ability of TBS alone or in combination with LS-BMD to predict osteoporotic fracture. It has been shown that TBS differentiates subjects with and without vertebral fractures (matched for age and BMD) for any BMD stratification [25,26,28]. Similar results have been recently reported for subjects with femoral neck fracture [27]. Moreover, the assessment of TBS in the evaluation of fracture risk in patients with secondary osteoporosis has also been reported: patients with subclinical hypercortisolism [30] or rheumatoid arthritis [29] with vertebral fractures had reduced TBS values. TBS better assessed fracture risk than lumbar spine BMD in these patients. On the other hand, the limitations of the TBS should also be mentioned. As extensively revised elsewhere [20], TBS does

not really measure bone micro-architecture. Moreover, it is difficult to discriminate, for instance, the contribution of trabecular texture in respect to that of bone geometry, soft tissue composition, possible artifacts due to osteoarthritis-related degenerative changes at the lumbar spine. The results of our study suggest that the measurement of TBS is more useful than that of BMD at the lumbar spine in the evaluation of vertebral fracture risk in postmenopausal women with PHPT. Moreover, the combination of both YSM and TBS reinforces the predicting value of TBS alone, possibly having an application in clinical practice to select high risk patients requiring a screening for vertebral fracture. The major limit of the current study lies in its cross-sectional design, which cannot conclude about causality between low TBS and occurrence of fractures. An ongoing longitudinal extension study from our group will probably help to clarify the predictive role of TBS for incident vertebral fracture. In this context, the only longitudinal data hitherto available, published by Eller-Vainicher and co-workers [30], show that in patients with subclinical hypercortisolism followed up for 24 months, TBS was predictive of incident vertebral fractures, regardless of LS-BMD, BMI, and age. These findings strongly support the potential clinical utility of TBS in the management of patients with bone disease affecting not only bone density but also bone quality. Due to its low cost and the easy application, TBS could represent an adjunctive tool to BMD testing in the assessment of fracture risk. In conclusion, TBS seems to indirectly reflect an alteration of bone micro-architecture in postmenopausal women with PHPT. In addition, TBS appears to be more accurate than BMD of the lumbar spine for identifying PHPT patients at risk for vertebral fractures. The combination of TBS and YSM would be helpful in clinical practice to select patients deserving a more complete radiological evaluation to detect non-symptomatic spine fractures. Contributors Study design: ER and SM. Study conduct: ER, CC, and SM. Data collection: CC, CC 1, MA, AD, JP, DD, and SP. Data analysis: ER, VC, and IN. Data interpretation: ER, VC, and SM. Drafting manuscript: ER and VC. Revising manuscript content: SM. Approving final version of manuscript: ER, CC, IN, CC 1, MA, AS, JP, DD, SP,VC, and SM. ER and SM take responsibility for the integrity of the data analysis. References [1] Pallan S, Rahman MO, Khan AA. Diagnosis and management of primary hyperparathyroidism. BMJ Mar 19 2012;344:e1013, http://dx.doi.org/10.1136/bmj.e1013. [2] Marcocci C, Cetani F. Clinical practice. Primary hyperparathyroidism. N Engl J Med 2011;365(25):2389–97. [3] Mosekilde L. Primary hyperparathyroidism and the skeleton. Clin Endocrinol Jul 2008;69(1):1–19. [4] Silverberg SJ, Shane E, de la Cruz L, Dempster DW, Feldman F, Seldin D, et al. Skeletal disease in primary hyperparathyroidism. J Bone Miner Res Jun 1989;4(3): 283–91. [5] Minisola S, Rosso R, Romagnoli E, Pepe J, De Geronimo S, Dionisi S, et al. Uneven deficits in vertebral bone density in postmenopausal patients with primary hyperparathyroidism as evaluated by posterior–anterior and lateral dual-energy absorptiometry. Osteoporos Int Aug 2002;13(8):618–23. [6] Dempster DW, Parisien M, Silverberg SJ, Liang X-G, Schnitzer M, Shen V, et al. On the mechanism of cancellous bone preservation in postmenopausal women with mild primary hyperparathyroidism. J Clin Endocrinol Metab May 1999;84(5): 1562–6. [7] Dempster DW, Müller R, Zhou H, Kohler T, Shane E, Parisien M, et al. Preserved three-dimensional cancellous bone structure in mild primary hyperparathyroidism. Bone Jul 2007;41(1):19–24. [8] Rubin MR, Cosman F, Lindsay R, Bilezikian JP. The anabolic effects of parathyroid hormone. Osteoporos Int 2002;13(4):267–77. 1

Claudia Castro.

E. Romagnoli et al. / Bone 53 (2013) 154–159 [9] Larsson K, Ljunghall S, Krusemo UB, Naessen T, Lindh E, Persson I. The risk of hip fractures in patients with primary hyperparathyroidism: a population-based cohort study with a follow-up of 19 years. J Intern Med Dec 1993;234(6):585–93. [10] Vestergaard P, Mollerup CL, Frøkjaer VG, Christiansen P, Blichert-Toft M, Mosekilde L. Cohort study of risk of fracture before and after surgery for primary hyperparathyroidism. BMJ 2000;321(7261):598–602. [11] Khosla S, Melton III LJ, Wermers RA, Crowson CS, O'Fallon WM, Riggs BL. Primary hyperparathyroidism and the risk of fracture: a population-based study. J Bone Miner Res 1999;14(10):1700–7. [12] Kenny AM, MacGillivray DC, Pilbeam CC, Crombie HD, Raisz LG. Fracture incidence in postmenopausal women with primary hyperparathyroidism. Surgery Jul 1995;118(1):109–14. [13] De Geronimo S, Romagnoli E, Diacinti D, D'Erasmo E, Minisola S. The risk of fractures in postmenopausal women with primary hyperparathyroidism. Eur J Endocrinol Sep 2006;155(3):415–20. [14] Vignali E, Viccica G, Diacinti D, Cetani F, Cianferotti L, Ambrogini E, et al. Morphometric vertebral fractures in postmenopausal women with primary hyperparathyroidism. J Clin Endocrinol Metab Jul 2009;94(7):2306–12. [15] Bilezikian JP. Bone strength in primary hyperparathyroidism. Osteoporos Int 2003;14(Suppl. 5):S113–5. [16] Link TM, Vieth V, Stehling C, Lotter A, Beer A, Newitt D, et al. High-resolution MRI vs multislice spiral CT: which technique depicts the trabecular bone structure best? Eur Radiol Apr 2003;13(4):663–71. [17] Patsch JM, Burghardt AJ, Kazakia G, Majumdar S. Noninvasive imaging of bone microarchitecture. Ann N Y Acad Sci Dec 2011;1240:77–87, http://dx.doi.org/ 10.1111/j.1749-6632.2011.06282.x. [18] Charopoulos I, Tournis S, Trovas G, Raptou P, Kaldrymides P, Skarandavos G, et al. Effect of primary hyperparathyroidism on volumetric bone mineral density and bone geometry assessed by peripheral quantitative computed tomography in postmenopausal women. J Clin Endocrinol Metab May 2006;91(5):1748–53. [19] Hansen S, Beck Jensen JE, Rasmussen L, Hauge EM, Brixen K. Effects on bone geometry, density, and microarchitecture in the distal radius but not the tibia in women with primary hyperparathyroidism: a case–control study using HR-pQCT. J Bone Miner Res Sep 2010;25(9):1941–7. [20] Bousson V, Bergot C, Sutter B, Levitz P, Cortet B, the Scientific Committee of the GRIO (Groupe de Recherche et d'Information sur les Ostéoporoses). Trabecular bone score (TBS): available knowledge, clinical relevance, and future prospects. Osteoporos Int May 2012;23(5):1489–501. [21] Pothuaud L, Carceller P, Hans D. Correlations between grey-level variations in 2D projection images (TBS) and 3D microarchitecture: applications in the study of human trabecular bone microarchitecture. Bone Apr 2008;42(4):775–87. [22] Pothuaud L, Benhamou CL, Porion P, Lespessailles E, Harba R, Levitz P. Fractal dimension of trabecular bone projection texture is related to three-dimensional microarchitecture. J Bone Miner Res Apr 2000;15(4):691–9. [23] Hans D, Barthe N, Boutroy S, Pothuaud L, Winzenrieth R, Krieg M-A. Correlations between trabecular bone score, measured using anteroposterior dual-energy X-ray absorptiometry acquisition, and 3-dimensional parameters of bone microarchitecture:

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32] [33]

[34]

[35]

[36]

159

an experimental study on human cadaver vertebrae. J Clin Densitom Jul-Sep 2011;14(3):302–12. Pothuaud L, Barthe N, Krieg MA, Mehsen N, Carceller P, Hans D. Evaluation of the potential use of trabecular bone score to complement bone mineral density in the diagnosis of osteoporosis: a preliminary spine BMD-matched, case–control study. J Clin Densitom Apr-Jun 2009;12(2):170–6. Rabier B, Héraud A, Grand-Lenoir C, Winzenrieth R, Hans D. A multicentre, retrospective case–control study assessing the role of trabecular bone score (TBS) in menopausal Caucasian women with low areal bone mineral density (BMDa): analysing the odds of vertebral fracture. Bone Jan 2010;46(1):176–81. Winzenrieth R, Dufour R, Pothuaud L, Hans D. A retrospective case–control study assessing the role of trabecular bone score in postmenopausal Caucasian women with osteopenia: analyzing the odds of vertebral fracture. Calcif Tissue Int Feb 2010;86(2):104–9. Del Rio LM, Winzenrieth R, Cormier C, Di Gregorio S. Is bone microarchitecture status of the lumbar spine assessed by TBS related to femoral neck fracture? A Spanish case–control study. Osteoporos Int May 12 2012 [Epub ahead of print]. Hans D, Goertzen A, Krieg MA, Leslie W. Bone microarchitecture assessed by TBS predicts hip, clinical spine and all osteoporotic fractures independently of BMD in 22234 women aged 50 and older: the Manitoba Prospective Study. J Bone Miner Res Nov 2011;26(11):2762–9, http://dx.doi.org/10.1002/jbmr.499. Breban S, Briot K, Kolta S, Paternotte S, Ghazi M, Fechtenbaum J, et al. Identification of rheumatoid arthritis patients with vertebral fractures using bone mineral density and trabecular bone score. J Clin Densitom Jul-Sep 2012;15(3):260–6. Eller-Vainicher C, Morelli V, Ulivieri FM, Palmieri S, Zhukouskaya VV, Cairoli E, et al. Bone quality, as measured by trabecular bone score (TBS), in patients with adrenal incidentalomas with and without subclinical hypercortisolism. J Bone Miner Res Oct 2012;27(10):2223–30, http://dx.doi.org/10.1002/jbmr.1648. Carnevale V, Manfredi G, Romagnoli E, De Geronimo S, Paglia F, Pepe J, et al. Vitamin D status in female patients with primary hyperparathyroidism: does it play a role in skeletal damage? Clin Endocrinol (Oxf) Jan 2004;60(1):81–6. Genant HK, Wu CY, van Knijk C, Nevitt M. Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res Sep 1993;8(9):1137–48. Cipriani C, Romagnoli E, Scarpiello A, Angelozzi M, Montesano T, Minisola S. Phalangeal quantitative ultrasound and bone mineral density in evaluating cortical bone loss: a study in postmenopausal women with primary hyperparathyroidism and subclinical iatrogenic hyperthyroidism. J Clin Densitom Oct-Dec 2009;12(4):456–60. Montagnani A, Gonnelli S, Cepollaro C, Bruni D, Franci MB, Lucani B, et al. Graphic trace analysis of ultrasound at the phalanges may differentiate between subjects with primary hyperparathyroidism and with osteoporosis: a pilot study. Osteoporos Int Mar 2002;13(3):222–7. Melton III LJ, Atkinson EJ, Michael O'Fallon W, Heath III H. Risk of age-related fractures in patients with primary hyperparathyroidism. Arch Intern Med Nov 1992;152:2269–73. Carnevale V, Pacitti MT, Pileri M, Paglia F, Scillitani A, Dionisi S, et al. Short-term effects of surgery in post-menopausal patients with primary hyperparathyroidism and normal bone turnover. J Endocrinol Invest Sep 2001;24(8):575–9.