Bone marrow perfusion measured with dynamic contrast enhanced magnetic resonance imaging is correlated to body mass index in adults

Bone 99 (2017) 47–52

Contents lists available at ScienceDirect

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

Full Length Article

Bone marrow perfusion measured with dynamic contrast enhanced magnetic resonance imaging is correlated to body mass index in adults Jean-François Budzik a,b,⁎, Guillaume Lefebvre c, Hélène Behal d, Sébastien Verclytte a, Pierre Hardouin e, Pedro Teixeira f, Anne Cotten a,e a

Lille Catholic Hospitals, Imaging Department, Lille Catholic University, Lille, France PMOI Physiopathology of Inflammatory Bone Diseases, EA 4490, Lille, France Lille Regional University Hospital, Musculoskeletal Imaging Department, University of Lille Nord de France, Lille, France d Lille Regional University Hospital, Biostatistics Department, University of Lille Nord de France, Lille, France e PMOI Physiopathology of Inflammatory Bone Diseases, EA 4490, University of Lille Nord de France, Lille, France f Nancy Regional University Hospital, Imaging Department, University of Lorraine, Nancy, France b c

a r t i c l e

i n f o

Article history: Received 6 October 2016 Revised 3 March 2017 Accepted 7 March 2017 Available online 24 March 2017 Keywords: Dynamic contrast-enhanced magnetic resonance imaging Bone marrow Vascularization Perfusion Body mass index Obesity

a b s t r a c t Bone marrow metabolism is complex and far from being fully understood. Novel aspects, such as the roles of bone marrow adiposity and vascularisation in bone metabolism currently attract attention. There is also a growing interest in the influence obesity might have on bone metabolism. Our objective was to determine the effect of BMI on bone marrow perfusion parameters using dynamic contrast-enhanced magnetic resonance imaging. This prospective monocentric study was approved by our local Ethics committee. Written consent was obtained. The right hip of 59 adults under 60 years old (mean age 37.5) was imaged with a dynamic 3D T1 spoiled gradient echo magnetic resonance imaging sequence. Mean BMI was 24.8 (+/−4.4). Perfusion parameters were measured in the acetabulum and femoral neck, in the greater trochanter, in the femoral head epiphysis and in the subcutaneous adipose tissue. Associations between perfusion parameters and BMI were studied using a linear mixed model adjusted for age and sex effects. Our results showed that as the BMI increased, the exchanges between blood and bone marrow appeared more important (increased Ktrans and Kep values, p = 0.018 and p = 0.002 respectively) and the intramedullary blood flow appeared increased (lower time to peak values, p = 0.0002). In the subcutaneous fat, as the BMI increased, the vascularization decreased (lower area under the curve and initial slope values, p = 0.019 and p = 0.013 respectively). These results suggest that there is a relation between bone marrow perfusion and BMI, and that subcutaneous fat and bone marrow fat have different microvascular behaviours. Researchers must be aware of the effect of BMI on bone marrow perfusion parameters when they build a MR research protocol and analyse their data. A better understanding of these findings may provide the basis for the management of obesity-related bone changes. © 2017 Published by Elsevier Inc.

1. Introduction In recent years there is a growing interest of researchers about bone marrow, especially bone marrow adiposity and bone marrow vascularization, and their possible involvement in bone metabolism [1,2]. Even if more and more data on bone marrow adiposity are gathered in the scientific literature, numerous questions on its physiological features remain. It is now well established that bone marrow fat is not a simple ⁎ Corresponding author at: Service d'imagerie médicale, Groupement des Hôpitaux de l'Institut Catholique de Lille, Hôpital St Vincent de Paul, Boulevard de Belfort, 59000 Lille, France. E-mail addresses: [email protected] (J.-F. Budzik), [email protected] (G. Lefebvre), [email protected] (H. Behal), [email protected] (S. Verclytte), [email protected] (P. Hardouin), [email protected] (P. Teixeira), [email protected] (A. Cotten).

http://dx.doi.org/10.1016/j.bone.2017.03.048 8756-3282/© 2017 Published by Elsevier Inc.

filling tissue, but a specific depot that is mainly present in yellow bone marrow and actively interacts with bone and hematopoietic cells [3]. This interplay between fat and the skeleton suggests a feedback system in which adipokines and other bone-derived molecules play a key role [4]. As these molecules transit through the blood vessels, we believe that a better knowledge of bone marrow microvascularization properties would certainly be helpful in the understanding of bone metabolism. Among the pathophysiological processes that affect bone marrow, there is a specific interest in the relationship existing between bone metabolism and obesity [2,4,5]. Especially, the role of obesity on bone health, for example in osteoporosis in which the protective role of obesity has recently come into question following clinical and epidemiological studies [4]. Many factors (behavioural, morphological, physiological…) could modulate the bone-fat crosstalk but their role is

48

J.-F. Budzik et al. / Bone 99 (2017) 47–52

insufficiently known [2,4]. A better understanding of this relationship could open new insights for the evaluation and management of patients with obesity related diseases. Dynamic Contrast-Enhanced MRI (DCE-MRI) is a technique that can assess bone marrow perfusion in humans in safe and minimally invasive conditions [6]. It can be performed on most clinical magnetic resonance scanners using a conventional dose of gadolinium-based contrast media. It can assess changes in tissue contrast media concentration in vivo yielding qualitative and quantitative data in form of time-concentration curves and numerical parameters, respectively that are correlated with various aspects of tissular perfusion. During a DCE-MRI acquisition, a time series of T1-weighted images are acquired with fast imaging techniques before, during and after the intravenous injection of a gadolinium-based contrast agent [6,7]. In each unit of volume explored with MRI (voxels), magnetic resonance signal intensity varies during the injection, as a reaction to the changing concentration of the contrast agent within the tissue. The resulting concentration-intensity curve is a reflection of these changes, and corresponds to a mix of vessels permeability, tissue perfusion and extravascular extra-cellular space [6]. Contrarily to conventional contrast-enhanced MR imaging, in which a single post-contrast acquisition is performed, in DCE-MRI, the analysis of this signal-intensity curve gives an insight into the microvascular properties of the tissue: parameters describe the progression of contrast agent concentration increase from baseline to the peak (initial slope, IS), the time needed to reach this peak (time to peak, TTP) and the total area under the curve (AUC) [6, 7]. To go further in the description of tissular physiology, other parameters can be derived from a pharmacokinetic model. Most often, the model is two-compartimental, considering that the contrast agent may be either in the vessels (plasma space) or in the extravascular extracellular space (Ve) [6]. In the commonly used Tofts model [8], the exchanges between these two compartments are defined as bidirectional and symmetrical. The transfer constant (Ktrans) and the rate constant (Kep) describe these exchanges. Using DCE-MRI, previous studies have shown that bone marrow perfusion decreases with age [9], and that an increasing bone marrow fat content corresponds to a decreased perfusion [10,11]. Thus, the distribution pattern of red and yellow bone marrow significantly influences bone marrow perfusion [11]. Women also have higher bone marrow perfusion than men [9]. In their DCE study of vertebral and pelvic bone marrow in 43 females, Breault et al. showed that Ktrans and Kep values were inversely correlated with age and fat fraction. AUC and Ve also decreased as fat fraction increased [10]. However, to the best of our knowledge, the relationship between bone marrow perfusion and BMI has never been studied. Indeed, among the consequences of obesity on bone, changes in bone marrow perfusion might exist. As BMI increases, changing interactions between fat and bone might involve modifications in bone marrow vascularization, possibly to convey fat- and bone-derived molecules. If such microvascular modifications exist, they might constitute a bias in further clinical studies. We hypothesize bone perfusion is influenced by BMI. Therefore, our objective was to determine the effect of BMI on bone marrow perfusion parameters using DCE-MRI.

examination of the lumbar spine or the pelvis. They were addressed for weak clinical suspicion of axial spondyloarthritis, but eventually did not fulfil the 2009 ASAS (assessment of spondyloarthritis international society) classification criteria. In order to avoid potential pathological changes of the bone marrow, we applied many non-inclusion criteria. The patients were not included if they reported: neoplastic, inflammatory, hematologic or rheumatologic diseases; osteoporosis or osteopenia; hyperparathyroidism, current acute or chronic inflammatory syndromes; history of hip osteoarthritis or osteonecrosis; current hip pain. Also we did not include patients with hip orthopaedic hardware, as it induces artefacts, or chronic renal failure, because gadoliniumbased contrast agents were used. Finally, the patients were not included if their hip joint and bones showed any abnormality on MR images. Clinical data collected were: age, gender and body mass index. 2.2. MRI protocol Patients were examined on a 3 Teslas MR scan (Ingenia, Philips Healthcare, The Netherlands). The hips were imaged with a coronal T1 spin echo sequence and short tau inversion recovery sequences acquired in axial and coronal planes. Three variable flip angles sequences (3°, 10° and 17°) were acquired before injection. A previously described dynamic 3D T1 spoiled gradient echo covered the right hip [12]. Its main features were as follows: 94 axial slices covered a field of view of 228 × 130 × 169 mm. Time of repetition, time of echo, flip angle and bandwidth per pixel were 4.5 and 2.1 ms, 10° and 389 Hz respectively. Acquisition and reconstruction matrix were 64 × 66 and 128 × 128 respectively. Temporal resolution was 13.5 s. Five baseline scans were acquired. At the beginning of the sixth scan, 0.1 mmol/kg of gadoteric acid (DOTAREM, Guerbet, France) were injected through a peripheral catheter positioned in an antecubital vein, with an automatic injector, at a rate of 2.5 ml/s followed by 20cm3 of saline flush. Twenty dynamic scans were collected. Total DCE examination time was 9 min. 2.3. Post-processing A senior musculoskeletal radiologist (J.F.B.), blinded to clinical data, analyzed DCE images with the open-source software Osirix [13] and DCE tool software (Kyung Sung, Body MRI research group, Stanford University, CA, USA), according to the methodology used in a previous feasibility study [12] and in a recent study [14]. Arterial Input Function was measured manually in the femoral artery and T1 maps were calculated. Tofts model was applied. As shown in Fig. 1, eight regions of interest (ROI) of 10 mm2 were drawn on DCE native images in areas of red and yellow bone marrow, identified by their aspect on T1-weighted images: acetabulum and femoral neck (red bone marrow), greater trochanter (extra-articular yellow bone marrow) and five in the femoral head epiphysis (yellow bone marrow; one in the center, four in the subchondral bone marrow). One ROI was positioned in the subcutaneous adipose tissue more than 1 cm away from the skin surface, to avoid chemical shift artefacts that were identified in our preliminary study [12]. For each ROI, semi-quantitative and pharmacokinetic parameters were calculated: initial slope (IS), area under the curve (AUC), time to peak (TTP), transfer constant (Ktrans), rate constant (Kep) and extra-vascular extra-cellular space volume (Ve).

2. Materials and methods 2.4. Statistical analysis This prospective monocentric study was approved by our local Ethics committee as an observational study. Oral and written information was delivered before the examination and written consent was obtained. 2.1. Population Sixty adults under sixty years old were included in this study. The subjects were referred to our imaging department for an MRI

Quantitative variables are expressed as mean (+/−standard deviation). Normality of distributions was checked graphically and using the Shapiro-Wilk test. Skewed distributed variables (IS, AUC, Ktrans, Kep and Ve) were log-transformed before analysis. Qualitative variables are expressed as number (percentage). Associations between each perfusion parameter and body mass index were studied using a linear mixed model in order to take into account the correlation between the repeated measures within subjects

J.-F. Budzik et al. / Bone 99 (2017) 47–52

49

Fig. 1. ROIs are shown on this T1-weighted magnetic resonance image acquired in the coronal plane. In the acetabulum and in the femoral neck (arrows), higher red bone marrow content is responsible for moderate signal hyperintensity compared to adjacent muscles. In areas of yellow bone marrow (femoral head epiphysis, arrowheads, and greater trochanter, thick arrow), the bone is clearly hyperintense compared to adjacent muscles. The signal intensity is close to that of subcutaneous fat (double arrowheads). Because a whole volume is imaged, all the ROIs cannot appear on a single coronal image. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(one measure per ROI) by including a random subject effect. These models were all adjusted for ROI, age and sex effects. We further investigated the homogeneity of the BMI associations with perfusion parameters between the studied ROIs by including the corresponding interaction term in the linear mixed models; the magnitude of the association in each ROI was estimated using linear contrasts. Statistical testing was done at the two-tailed α level of 0.05. Data were analyzed using the SAS software package, release 9.3 (SAS Institute, Cary, NC).

3. Results DCE-MRI was performed successfully in 59 of the 60 subjects included. One was excluded because motion artefacts impaired DCE analysis. 31 patients were women and 28 were men. Mean age was 37.5 (± 12.5) and mean BMI was 24.8 (± 4.4). The age ranged from18 to 60 years and BMI ranged from 17.0 to 36.3. In red and yellow bone marrow, BMI was negatively correlated with TTP whereas it was positively correlated with Ktrans and Kep. The influence on TTP was found in each zone, whereas the effect of BMI on Kep and Ktrans was found in respectively 8 in 9 zones and 6 in 9 zones. BMI did not influence significantly IS and Ve values in bone marrow. In subcutaneous fat, BMI was negatively correlated with AUC (p = 0.019) and IS (p = 0,013). BMI did not influence significantly the other perfusion parameters evaluated in the subcutaneous tissue.

The values of the perfusion parameters and the correlation of these values with BMI are detailed in Tables 1 and 2.

4. Discussion BMI appears to have an influence on hip bone marrow perfusion parameters. As BMI increased, the exchanges between blood and bone marrow appeared more important (increased Ktrans and Kep values) and the intramedullary blood flow appeared increased (lower TTP values). To the best of our knowledge, there is no study reporting correlation of bone marrow perfusion measured in vivo and BMI, either in humans or in animals, to this day. Therefore, we cannot compare our results with other similar study. In our preliminary study, we found that Ktrans, Kep, IS and AUC values were different between red and yellow bone marrow [12]. Yet, the analysis was limited to 21 patients and we did not calculate patients' BMI. In our current study, BMI was correlated with DCE parameters in red and yellow bone marrow (analyzed separately), whatever the differences existing between them. In their DCE-MRI study of 43 female lumbar spine and pelvis, Breault et al. [10] did not take BMI into consideration. They reported an influence of age and fat fraction, which were negatively correlated with bone marrow perfusion. In our statistical analysis, we adjusted the models for age to avoid bias. It is important to note that areas of yellow

50

J.-F. Budzik et al. / Bone 99 (2017) 47–52

Table 1 Association between bone marrow perfusion parameters and body mass index, overall and according to ROI. Red bone marrow

Yellow bone marrow Femoral head

IS

AUC

TTP

Ktrans

Kep

Ve

Estimate ± SE p Estimate ± SE p Estimate ± SE p Estimate ± SE p Estimate ± SE p Estimate ± SE p

Greater trochanter

Subcutaneous fat

Overalla

P for heterogeneityb

0.02 ± 0.03 0.51 0.02 ± 0.01 0.17 −4.3 ± 1.1 0.0002 0.06 ± 0.02 0.018 0.05 ± 0.02 0.002 0.01 ± 0.02 0.63

0.02

Acetabulum Femoral neck

Superior

Center

Anterior

Posterior

Inferior

0.04 ± 0.02

0.006 ± 0.05 0.91 0.02 ± 0.02 0.31 −4.82 ± 2.2 0.029 0.09 ± 0.03 0.010 0.09 ± 0.03 0.004 0.003 ± 0.02 0.88

0.08 ± 0.05 0.07 0.03 ± 0.02 0.044 −5.4 ± 1.9 0.007 0.08 ± 0.03 0.011 0.07 ± 0.03 0.019 0.02 ± 0.02 0.39

0.06 ± 0.05 0.19 0.03 ± 0.02 0.07 −4.3 ± 2.0 0.041 0.11 ± 0.03 0.001 0.09 ± 0.03 0.001 0.02 ± 0.02 0.51

0.07 ± 0.06 0.21 0.04 ± 0.02 0.038 −5.1 ± 2.0 0.014 0.11 ± 0.03 0.002 0.08 ± 0.02 0.001 0.02 ± 0.02 0.33

0.04 ± 0.06

−0.02 ± 0.06

−0.12 ± 0.05

0.46 −0.0002 ± 0.02 0.99 −4.8 ± 1.9

0.63 0.02 ± 0.02

0.013 −0.04 ± 0.01

0.17 −5.5 ± 2.2

0.019 2.7 ± 2.5

0.012 0.01 ± 0.03

0.015 0.06 ± 0.03

0.30 −0.05 ± 0.04

0.68 0.008 ± 0.02

0.10 0.07 ± 0.03

0.15 −0.04 ± 0.03

0.73 0.003 ± 0.02

0.044 −0.01 ± 0.02

0.25 −0.02 ± 0.03

0.60

0.60

0.88

0.13 0.02 ± 0.01 0.14 −4.05 ± 2.0 0.047 0.06 ± 0.03 0.051 0.04 ± 0.02 0.045 0.03 ± 0.03 0.41

0.008 ± 0.03 0.76 0.01 ± 0.02 0.48 −6.94 ± 2.7 0.012 0.06 ± 0.03 0.046 0.05 ± 0.02 0.014 0.02 ± 0.02 0.45

0.04

0.21

b0.0001

0.0002

0.54

Linear regression coefficients (“estimate”), standard errors and p-values were adjusted for age and sex. Significant p-values are in bold characters. a The overall p value reflects the global effect of BMI on each perfusion parameter, without taking the zones into consideration. b For a given perfusion parameter, “p for heterogeneity” shows whether the effect of BMI is significantly different between the zones or not.

bone marrow were not included in the DCE analysis made by Breault et al. On the contrary, our analysis was made in areas of red and yellow bone marrow, and the influence of BMI on Ktrans, Kep and TTP values was observed in both types of marrow (Table 1). The comparison with other former DCE studies of healthy volunteers is limited, because none took BMI into consideration [15–19]. To the best of our knowledge, no study compared bone marrow composition between obese and non-obese subjects. Bredella et al. reported that vertebral bone marrow fat is positively correlated with visceral adipose tissue [20]. If we would extrapolate these results to hip bone marrow, we might assume that bone marrow fat content increases with BMI, and that fewer vessels might be present within the bone. This might be consistent with our results, as the parameters that are influenced by BMI in our study reflect the intramedullary blood flow and the exchanges between bone and vessels. Indeed, even if the number of vessels would decrease, adaptive phenomenon might result in accelerating the blood flow or increasing the trans-parietal exchanges occurring between vessels and bone cells via local hormonal regulation. DCEMRI does not have the ability to image microvessels within bone. Moreover, as far as we know, no imaging modality has the ability to image bone microvessels in vivo in humans. The role of trabecular bone volume is also a concern. Former studies using dual-energy X-ray absorptiometry (DXA) demonstrated that patients with higher BMI values had decreased trabecular bone scores [21– 23]. Although this parameter reflects bone trabecular microarchitecture,

one must keep in mind that is not a direct measurement, essentially because of technical factors [24]. More detailed analysis of bone microstructure is feasible with quantitative computerized tomography (QCT). Using DXA and QCT measurements in the radius and the tibia, Evans et al. demonstrated that obese adults have thicker and denser cortical bone, and more bone trabeculae than normal adults [25]. In a QCT study of the proximal hip in 3067 men, Shen et al. reported that for non-obese men (BMI b 30), BMI was positively associated with integral, cortical and trabecular volumetric bone mineral density, integral volume, cross-sectional area, and cortical volume [26]. However, for obese men (BMI ≥ 30), they found no association between increasing BMI and any of those parameters. Consequently, as trabecular volume and bone marrow fat increase in obese patients, we can hypothesize that the volume of hematopoietic bone marrow decreases. One could expect a decrease in bone marrow perfusion. We can hypothesize that adaptive phenomenon occur to balance such a decrease. This could be consistent with our results that suggest that the intramedullary blood flow and the exchanges between bone marrow and bone vessels increase with BMI. Yet, we could not assess the potential role of bone microstructure because we did not perform DXA or QCT. Thereafter, we discuss several hypotheses that might explain our results. First, a bone marrow conversion (from red to yellow marrow), which is a physiological phenomenon occurring during childhood

Table 2 Perfusion parameters values in each ROI. Values are expressed as mean ± SD. Red bone marrow

Yellow bone marrow

Subcutaneous fat

Femoral head

IS AUC TTP Ktrans Kep Ve

Greater trochanter

Acetabulum

Femoral neck

Superior

Anterior

Center

Posterior

Inferior

6.0 ± 4.2 34.9 ± 17.9 336.9 ± 62.0 0.27 ± 0.23 1.4 ± 1.2 0.22 ± 0.18

3.8 ± 2.8 25.1 ± 14.7 401.5 ± 85.3 0.16 ± 0.14 0.9 ± 0.6 0.18 ± 0.15

0.3 ± 0.5 4.7 ± 3.3 474.2 ± 67.7 0.02 ± 0.02 0.6 ± 0.4 0.04 ± 0.04

0.4 ± 0.5 5.5 ± 3.8 451.7 ± 64.6 0.02 ± 0.03 0.7 ± 0.5 0.04 ± 0.04

0.3 ± 0.4 5.7 ± 3.1 489.0 ± 62.9 0.02 ± 0.02 0.5 ± 0.4 0.05 ± 0.04

0.7 ± 2.8 5.8 ± 4.6 460.1 ± 67.3 0.03 ± 0.04 0.5 ± 0.3 0.05 ± 0.06

0.4 ± 0.6 6.7 ± 4.1 475.9 ± 59.2 0.02 ± 0.02 0.4 ± 0.2 0.05 ± 0.05

0.2 ± 0.3 3.8 ± 2.0 479.5 ± 70.5 0.02 ± 0.03 0.5 ± 0.5 0.04 ± 0.07

0.3 ± 0.4 5.6 ± 2.7 495.1 ± 78.8 0.03 ± 0.03 0.5 ± 0.4 0.06 ± 0.08

J.-F. Budzik et al. / Bone 99 (2017) 47–52

[19], can be discussed. Indeed, an increase in the red marrow content has been reported in obese patients on MR imaging [27]. However, a recent DCE-MRI study on hip bone marrow, using the same perfusion parameters evaluated herein, demonstrated that all parameters differed between red and yellow bone marrow [12]. As BMI had no influence on IS, AUC or Ve values in our study, thus we believe that BMI-related marrow perfusion changes cannot be explained only by changes in the proportion of red/yellow marrow. This is consistent with recent review articles that do not describe such obesity related bone marrow conversion [4,28]. Secondly, mechanical phenomena might play a role. Because the hip is a weight-bearing joint, an increase in body weight results in an increased pressure on the femoral head. In a clinical study, Beck et al. reported a link between hip intra-articular pressure and the blood supply to the femoral head [29]. During surgical procedures, intra-articular saline injection (10 to 35 ml) resulted in reduced blood flow to the femoral head measured with laser Doppler flowmetry directly in the bone. In our study, intramedullary blood flow appeared increased. So we might hypothesize that mechanical factors could affect blood supply and explain that changes in perfusion parameters were found in all the studied zones. However it is important to notice that the experimental conditions of Beck et al.'s were far from physiological conditions. Thirdly, metabolic changes can be hypothesized. Vertebral bone marrow fat is positively correlated with visceral bone marrow fat [20] and is acutely responsive to changes in systemic metabolism [28]. Adipocytes interact with their local environment by secreting adipokines, like leptin and adiponectin [30,31], and increase in size and number into the bone marrow spaces in obesity [32]. Moreover, brown adipose tissue is increased in obesity and has a rich vascular network [33]. Obesity related increase in brown adipose tissue could also influence bone marrow perfusion as more vessels might imply more exchanges with bone marrow. Such metabolic changes might explain why perfusion parameters were influenced similarly by BMI in all the zones studied. So, metabolic changes are likely, and a mechanical influence is possible. However, without microvessel density evaluation and precise quantification of bone adipocytes and brown adipose tissue, the proposed hypotheses remain speculative. We also found a correlation between BMI and the perfusion parameters in subcutaneous fat. As BMI increased, the perfusion of subcutaneous fat decreased (lower IS and AUC values), indicating that BMI influence was different between subcutaneous fat and bone marrow. This was particularly true in yellow bone marrow, which is mainly constituted of bone marrow fat. It is known that subcutaneous adipocytes and bone marrow adipocytes have phenotypic differences [28] and have different responses to systemic stimuli [2,28]. These results suggest that subcutaneous fat and bone marrow fat are likely to have also different microvascular behaviours. The linear regression coefficients that we found are small. The question of the potential clinical significance of these variations cannot be eluded. DCE-MRI is essentially a research tool in musculoskeletal imaging. As it is not a routine tool in non-tumoral bone diseases imaging, we cannot say whether it is clinically significant or not. In bone tumoral diseases, DCE can be used for the diagnosis and follow-up of osteoid osteomas [34], for the diagnosis of cartilaginous bone tumors [35] and for the pre-therapeutic assessment of non-union fractures [36]. In these situations, DCE is used to characterize an intensely increased bone vascularization in pathological bone and adjacent muscle is used as a reference tissue. So changes occurring in the normal bone cannot have consequences on DCE analysis. On the contrary, DCE-MRI can be easily used in clinical research protocols. Many parameters may influence bone marrow vascularization. Age, sex, and bone marrow fat content have already been identified as such [9,10]. Now, BMI is another parameter to take into account when analysing bone perfusion. As sources of artefacts are numerous in MRI practice, the potential problem of technical limitations has to be evoked. Two main kinds of

51

MRI artefacts might have biased the results. First, chemical shift artefacts could occur in the subcutaneous fat that lies just under the skin. Despite the use of very short echo times (2.1 ms), such artefacts were identified in our preliminary study [12]. These artefacts occurred in the right/left direction and were limited to one or two pixels. Because we identified this potential problem, the ROIs used to measure subcutaneous fat perfusion parameters were drawn more than 1 cm away from the skin surface. Moreover, in patients with higher BMI, and thus potential higher waist circumference, the deepness of the hip could be increased. The signal measured by the phased array coils might then have been decreased. However, in our preliminary study of 21 patients, we measured signalto-noise and contrast-to-noise ratios in deep structures (cortical and medullary bone from the femoral diaphysis). These ratios were very high (respectively 203 and 176 for pre-contrast measurements), showing that no significant signal loss existed in these areas. Our study is limited by the absence of bone tissue histologic analysis confirming perfusion findings. However, previous studies indicated that MRI is a valid tool for the non-invasive assessment of bone marrow and tissue perfusion [1,2,6,10]. Also, we did not consider life habits (such as smoking, alcohol consumption, sports practice, sedentary lifestyle) and previous history of cardiovascular diseases, high blood pressure, and hypercholesterolemia. Further longitudinal studies comparing obese and non-obese patients with control of metabolic parameters and hormonal levels are required to better understand the involvement of bone marrow perfusional changes in physiopathology. Also, we could not measure the proportion of bone marrow fat. The identification of red and yellow marrow was based on bone marrow signal intensity on T1-weighted sequences. This method is used as a daily basis in clinical work by musculoskeletal radiologists [37,38]. However, there is no proportional relational between the signal intensity on T1weighted sequences and the proportion of bone marrow fat because other technical factors interfere. In order to measure bone marrow adiposity, other sequences could have been done: chemical shift imaging [39] or MR spectroscopy [40]. But we did not include such sequences in our MR protocol because they were not necessary to test our hypothesis. As a conclusion, this study reports a correlation between BMI and bone marrow perfusion measured in vivo in humans with DCE-MRI: the intramedullary blood flow and the exchanges between bone marrow and bone vessels increase with BMI. Researchers must be aware of the effect of BMI on bone marrow perfusion parameters when they build a MR research protocol and analyse their data. Our results also suggest that subcutaneous fat and bone marrow fat have different microvascular behaviours. This information provides further insight on bone marrow physiopathology and should be considered when evaluating marrow perfusion on MRI. Authors' roles Study design: JFB, AC, PT, SV, HB. Study conduct: JFB. Data collection: JFB, GL. Data analysis: JFB. Data interpretation: JFB, AC, PT, SV, HB, PH. Drafting manuscript: JFB, HB. Revising manuscript content: PT, AC, SV, PH. Approving final version of manuscript: JFB, GL, AC, PT, SV, HB, PH. AC takes responsibility for the integrity of the data analysis. Grant support The authors did not benefit from any grant or funding. Acknowledgments No acknowledgments.

52

J.-F. Budzik et al. / Bone 99 (2017) 47–52

References [1] P. Hardouin, P.J. Marie, C.J. Rosen, New insights into bone marrow adipocytes: report from the First European Meeting on Bone Marrow Adiposity (BMA 2015), Bone S8756-3282 (2015) 414–417, http://dx.doi.org/10.1016/j.bone.2015.11.013. [2] J. Paccou, P. Hardouin, A. Cotten, G. Penel, B. Cortet, The role of bone marrow fat in skeletal health: usefulness and perspectives for clinicians, J. Clin. Endocrinol. Metab. 100 (2015) 3613–3621, http://dx.doi.org/10.1210/jc.2015-2338. [3] P. Hardouin, V. Pansini, B. Cortet, Bone marrow fat, Jt. Bone Spine Rev. Rhum. 81 (2014) 313–319, http://dx.doi.org/10.1016/j.jbspin.2014.02.013. [4] E.A. Greco, A. Lenzi, S. Migliaccio, The obesity of bone, Ther. Adv. Endocrinol. Metab. 6 (2015) 273–286, http://dx.doi.org/10.1177/2042018815611004. [5] A.L. Evans, M.A. Paggiosi, R. Eastell, J.S. Walsh, Bone density, microstructure and strength in obese and normal weight men and women in younger and older adulthood, J. Bone Miner. Res. Off. J. Am. Soc. Bone Miner. Res. 30 (2015) 920–928, http:// dx.doi.org/10.1002/jbmr.2407. [6] M. Essig, M.S. Shiroishi, T.B. Nguyen, M. Saake, J.M. Provenzale, D. Enterline, N. Anzalone, A. Dörfler, A. Rovira, M. Wintermark, M. Law, Perfusion MRI: the five most frequently asked technical questions, AJR Am. J. Roentgenol. 200 (2013) 24–34, http://dx.doi.org/10.2214/AJR.12.9543. [7] F. Khalifa, A. Soliman, A. El-Baz, M. Abou El-Ghar, T. El-Diasty, G. Gimel'farb, R. Ouseph, A.C. Dwyer, Models and methods for analyzing DCE-MRI: a review, Med. Phys. 41 (2014) 124301, http://dx.doi.org/10.1118/1.4898202. [8] P.S. Tofts, G. Brix, D.L. Buckley, J.L. Evelhoch, E. Henderson, M.V. Knopp, H.B. Larsson, T.Y. Lee, N.A. Mayr, G.J. Parker, R.E. Port, J. Taylor, R.M. Weisskoff, Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols, J. Magn. Reson. Imaging JMRI. 10 (1999) 223–232. [9] W.T. Chen, T.T. Shih, R.C. Chen, S.Y. Lo, C.T. Chou, J.M. Lee, H.Y. Tu, Vertebral bone marrow perfusion evaluated with dynamic contrast-enhanced MR imaging: significance of aging and sex, Radiology 220 (2001) 213–218. [10] S.R. Breault, T. Heye, M.R. Bashir, B.M. Dale, E.M. Merkle, C.S. Reiner, K.F. Faridi, R.T. Gupta, Quantitative dynamic contrast-enhanced MRI of pelvic and lumbar bone marrow: effect of age and marrow fat content on pharmacokinetic parameter values, AJR Am. J. Roentgenol. 200 (2013) W297–W303, http://dx.doi.org/10.2214/ AJR.12.9080. [11] A. Biffar, O. Dietrich, S. Sourbron, H.-R. Duerr, M.F. Reiser, A. Baur-Melnyk, Diffusion and perfusion imaging of bone marrow, Eur. J. Radiol. 76 (2010) 323–328, http://dx. doi.org/10.1016/j.ejrad.2010.03.011. [12] J.-F. Budzik, G. Lefebvre, G. Forzy, M. El Rafei, D. Chechin, A. Cotten, Study of proximal femoral bone perfusion with 3D T1 dynamic contrast-enhanced MRI: a feasibility study, Eur. Radiol. 24 (2014) 3217–3223, http://dx.doi.org/10.1007/s00330-0143340-5. [13] A. Rosset, L. Spadola, O. Ratib, OsiriX: an open-source software for navigating in multidimensional DICOM images, J. Digit. Imaging 17 (2004) 205–216, http://dx.doi. org/10.1007/s10278-004-1014-6. [14] J.-F. Budzik, J. Ding, L. Norberciak, T. Pascart, H. Toumi, S. Verclytte, R. Coursier, Perfusion of subchondral bone marrow in knee osteoarthritis: a dynamic contrast enhanced magnetic resonance imaging preliminary study, Eur J Radiol. (2016)http:// dx.doi.org/10.1016/j.ejrad.2016.12.023. [15] G.M. Müller, S. Månsson, M.F. Müller, O. Ekberg, A. Björkman, Assessment of perfusion in normal carpal bones with dynamic gadolinium-enhanced MRI at 3 Tesla, J. Magn. Reson. Imaging JMRI. 38 (2013) 168–172, http://dx.doi.org/10.1002/jmri. 23951. [16] H.T. Ma, H. Lv, J.F. Griffith, A.F.W. Li, D.K.W. Yeung, J. Leung, P.-C. Leung, J. Yuan, Perfusion and bone mineral density as function of vertebral level at lumbar spine, Conf. Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. IEEE Eng. Med. Biol. Soc. Conf. 2012 (2012) 3488–3491, http://dx.doi.org/10.1109/EMBC.2012.6346717. [17] J. Hillengass, B. Stieltjes, T. Bäuerle, F. McClanahan, C. Heiss, T. Hielscher, B. WagnerGund, V. Habetler, H. Goldschmidt, H.-P. Schlemmer, S. Delorme, C.M. Zechmann, Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and diffusion-weighted imaging of bone marrow in healthy individuals, Acta Radiol. Stockh. Swed. 1987 (52) (2011) 324–330, http://dx.doi.org/10.1258/ar.2010.100366. [18] V. Savvopoulou, T.G. Maris, L. Vlahos, L.A. Moulopoulos, Differences in perfusion parameters between upper and lower lumbar vertebral segments with dynamic contrast-enhanced MRI (DCE MRI), Eur. Radiol. 18 (2008) 1876–1883, http://dx.doi. org/10.1007/s00330-008-0943-8. [19] M.A. Bedoya, C. Jaimes, D′mitry Khrichenko, J. Delgado, B.J. Dardzinski, D. Jaramillo, Dynamic gadolinium-enhanced MRI of the proximal femur: preliminary experience in healthy children, AJR Am. J. Roentgenol. 203 (2014) W440-446, http://dx.doi.org/ 10.2214/AJR.13.12341. [20] M.A. Bredella, M. Torriani, R.H. Ghomi, B.J. Thomas, D.J. Brick, A.V. Gerweck, C.J. Rosen, A. Klibanski, K.K. Miller, Vertebral bone marrow fat is positively associated with visceral fat and inversely associated with IGF-1 in obese women, Obes. Silver Spring Md. 19 (2011) 49–53, http://dx.doi.org/10.1038/oby.2010.106.

[21] M. Schorr, L.E. Dichtel, A.V. Gerweck, M. Torriani, K.K. Miller, M.A. Bredella, Body composition predictors of skeletal integrity in obesity, Skelet. Radiol. 45 (2016) 813–819, http://dx.doi.org/10.1007/s00256-016-2363-1. [22] E. Romagnoli, C. Lubrano, V. Carnevale, D. Costantini, L. Nieddu, S. Morano, S. Migliaccio, L. Gnessi, A. Lenzi, Assessment of trabecular bone score (TBS) in overweight/obese men: effect of metabolic and anthropometric factors, Endocrine 54 (2016) 342–347, http://dx.doi.org/10.1007/s12020-016-0857-1. [23] L. Langsetmo, T.N. Vo, K.E. Ensrud, B.C. Taylor, P.M. Cawthon, A.V. Schwartz, D.C. Bauer, E.S. Orwoll, N.E. Lane, E. Barrett-Connor, J.T. Schousboe, Osteoporotic fractures in men (MrOS) research group, the association between trabecular bone score and lumbar spine volumetric BMD is attenuated among older men with high body mass index, J. Bone Miner. Res. Off. J. Am. Soc. Bone Miner. Res. 31 (2016) 1820–1826, http://dx.doi.org/10.1002/jbmr.2867. [24] V. Bousson, C. Bergot, B. Sutter, P. Levitz, B. Cortet, Scientific Committee of the Groupe de Recherche et d'Information sur les Ostéoporoses, Trabecular bone score (TBS): available knowledge, clinical relevance, and future prospects, Osteoporos. Int. J. Establ. Result Coop. Eur. Found. Osteoporos. Natl. Osteoporos. Found. USA. 23 (2012) 1489–1501, http://dx.doi.org/10.1007/s00198-011-1824-6. [25] A.L. Evans, M.A. Paggiosi, R. Eastell, J.S. Walsh, Bone density, microstructure and strength in obese and normal weight men and women in younger and older adulthood, J. Bone Miner. Res. Off. J. Am. Soc. Bone Miner. Res. 30 (2015) 920–928, http:// dx.doi.org/10.1002/jbmr.2407. [26] J. Shen, C.M. Nielson, L.M. Marshall, D.C. Lee, T.M. Keaveny, E.S. Orwoll, Osteoporotic fractures in men MrOS research group, the association between BMI and QCT-derived proximal hip structure and strength in older men: a cross-sectional study, J. Bone Miner. Res. Off. J. Am. Soc. Bone Miner. Res. 30 (2015) 1301–1308, http://dx. doi.org/10.1002/jbmr.2450. [27] T.B. Poulton, W.D. Murphy, J.L. Duerk, C.C. Chapek, D.H. Feiglin, Bone marrow reconversion in adults who are smokers: MR imaging findings, AJR Am. J. Roentgenol. 161 (1993) 1217–1221, http://dx.doi.org/10.2214/ajr.161.6.8249729. [28] E.L. Scheller, C.J. Rosen, What's the matter with MAT? Marrow adipose tissue, metabolism, and skeletal health, Ann. N. Y. Acad. Sci. 1311 (2014) 14–30, http://dx. doi.org/10.1111/nyas.12327. [29] M. Beck, K.A. Siebenrock, B. Affolter, H. Nötzli, J. Parvizi, R. Ganz, Increased intraarticular pressure reduces blood flow to the femoral head, Clin. Orthop. (2004) 149–152. [30] P. Hardouin, V. Pansini, B. Cortet, Bone marrow fat, Joint Bone Spine. 81 (2014) 313–319, http://dx.doi.org/10.1016/j.jbspin.2014.02.013. [31] K.J. Suchacki, W.P. Cawthorn, C.J. Rosen, Bone marrow adipose tissue: formation, function and regulation, Curr. Opin. Pharmacol. 28 (2016) 50–56, http://dx.doi. org/10.1016/j.coph.2016.03.001. [32] B.J. Adler, K. Kaushansky, C.T. Rubin, Obesity-driven disruption of haematopoiesis and the bone marrow niche, Nat. Rev. Endocrinol. 10 (2014) 737–748, http://dx. doi.org/10.1038/nrendo.2014.169. [33] S.C. Sampath, S.C. Sampath, M.A. Bredella, A.M. Cypess, M. Torriani, Imaging of brown adipose tissue: state of the art, Radiology 280 (2016) 4–19, http://dx.doi. org/10.1148/radiol.2016150390. [34] P.A.G. Teixeira, A. Chanson, M. Beaumont, S. Lecocq, M. Louis, B. Marie, F. Sirveaux, A. Blum, Dynamic MR imaging of osteoid osteomas: correlation of semiquantitative and quantitative perfusion parameters with patient symptoms and treatment outcome, Eur. Radiol. 23 (2013) 2602–2611, http://dx.doi.org/10.1007/s00330-0132867-1. [35] T. De Coninck, L. Jans, G. Sys, W. Huysse, T. Verstraeten, R. Forsyth, B. Poffyn, K. Verstraete, Dynamic contrast-enhanced MR imaging for differentiation between enchondroma and chondrosarcoma, Eur. Radiol. 23 (2013) 3140–3152, http://dx. doi.org/10.1007/s00330-013-2913-z. [36] O. Schoierer, K. Bloess, D. Bender, I. Burkholder, H.-U. Kauczor, G. Schmidmaier, M.A. Weber, Dynamic contrast-enhanced magnetic resonance imaging can assess vascularity within fracture non-unions and predicts good outcome, Eur. Radiol. 24 (2014) 449–459, http://dx.doi.org/10.1007/s00330-013-3043-3. [37] B.Y. Chan, K.G. Gill, S.L. Rebsamen, J.C. Nguyen, MR imaging of pediatric bone marrow, Radiogr. Rev. Publ. Radiol. Soc. N. Am. Inc. 36 (2016) 1911–1930, http://dx. doi.org/10.1148/rg.2016160056. [38] L.M. Shah, C.J. Hanrahan, MRI of spinal bone marrow: part I, techniques and normal age-related appearances, AJR Am. J. Roentgenol. 197 (2011) 1298–1308, http://dx. doi.org/10.2214/AJR.11.7005. [39] M. Winfeld, S. Ahlawat, N. Safdar, Utilization of chemical shift MRI in the diagnosis of disorders affecting pediatric bone marrow, Skelet. Radiol. 45 (2016) 1205–1212, http://dx.doi.org/10.1007/s00256-016-2403-x. [40] V. Pansini, A. Monnet, J. Salleron, P. Hardouin, B. Cortet, A. Cotten, 3 Tesla (1) H MR spectroscopy of hip bone marrow in a healthy population, assessment of normal fat content values and influence of age and sex, J. Magn. Reson. Imaging JMRI. 39 (2014) 369–376, http://dx.doi.org/10.1002/jmri.24176.