- Email: [email protected]
Correlations between strength and quantitative computed tomography measurement of callus mineralization in experimental tibial fractures
Clinical Biomechanics 26 (2011) 95–100
Contents lists available at ScienceDirect
Clinical Biomechanics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c l i n b i o m e c h
Correlations between strength and quantitative computed tomography measurement of callus mineralization in experimental tibial fractures Ulf Sigurdsen a,b,⁎, Olav Reikeras c, Arne Hoiseth d, Stein Erik Utvag e,a a
Department of Orthopedic Surgery, University of Oslo, Norway Institute of Surgical Research, Oslo University Hospital Rikshospitalet, Norway Department of Orthopedic Surgery, Oslo University Hospital Rikshospitalet, Norway d Curato Roentgen Inc., Oslo, Norway e Department of Orthopedic Surgery, Akershus University Hospital, Norway b c
a r t i c l e
i n f o
Article history: Received 21 April 2010 Accepted 8 September 2010 Keywords: Experimental fracture healing Tibial diaphyseal fracture Quantitative computed tomography External fixation Intramedullary nailing
a b s t r a c t Background: The evaluation of fracture healing in the clinic has not changed significantly during the past few decades, despite the development of modern tissue-imaging tools. Recent publications have reported significant and interesting associations between biomechanical properties and quantitative computed tomography data of fractures and grafts. We therefore studied the correlations between the strength and segmented quantitative computed tomography data of tibial diaphyseal fractures. Methods: Forty male rats received a tibial-shaft osteotomy that was initially stabilized with either intramedullary nailing or external fixation. Evaluation at 30 and 60 days post-osteotomy included X-ray, quantitative computed tomography and bending testing. Quantitative computed tomography data were segmented by voxel density into soft callus (171–539 mg/cm3), hard callus (540–1199 mg/cm3) and cortical bone (≥ 1200 mg/cm3), and volumetric bone mineral density was calculated. Findings: All fractures demonstrated pronounced formation of soft and hard callus tissues at 30 days postosteotomy, and at 60 days the cortical bone volume was significantly increased with callus resorption. Bending strength correlated significantly and positively with fracture-site cortical bone volume and volumetric bone mineral density in the intramedullary nailed group in the early phase of healing. Interpretation: Quantitative computed tomography was used to quantify characteristic secondary healing. The observed correlations indicate that biomechanically important mineralization can be measured by quantitative computed tomography in the early phase of healing in flexibly fixed fractures. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Over the past few decades, fracture healing has been evaluated by patient interview, clinical examination and standard biplane radiograms. The dramatic developments in advanced tissue-imaging techniques and recent advances within computational science have not yet provided the orthopaedic surgeon with non-invasive methods that are more accurate at evaluating bone repair. Several authors have attributed this to the normally good fracture prognosis. However, an increasing number of patients present with co-morbidities that impair bone quality and repair, such as age-related osteoporosis and diabetes. There is also an increased demand for accurate staging of fracture healing for the early and safe removal of restrictions and surgical implants. Tibial
⁎ Corresponding author. Institute of Surgical Research, Oslo University Hospital Rikshospitalet, 0027 Oslo, Norway. E-mail address: [email protected] (U. Sigurdsen). 0268-0033/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2010.09.004
diaphyseal fractures can be among the most difficult fractures to treat, and the surgeon's choice is often between intramedullary nailing (IMN) and external fixation (EF) (Court-Brown, 2006). Measurement of the mineralization of intact bone provides a valuable estimation of bone quantity and quality, and future fracture risk (Markel et al., 1990). We know that exceptional biomechanical properties such as toughness characterize bone tissue, and that these properties are associated mainly with the amount and histological organization of calcium minerals. Other biomechanically relevant, but less-measurable tissue factors include bone microstructure, mineral crystal size, and the number and size of bone pores (Powell et al., 1989). Intact bone mineralization measured by several techniques is statistically correlated with biomechanical properties and clinically valuable fracture risk predictors (Markel et al., 1990). Measurement of bone mass—either as bone mineral density (BMD; area density in grams per square centimetre) or bone mineral content (in grams)—by dual X-ray absorptiometry is one such method for quantitative bone assessment. BMD combined with standard age–gender tables is a preferred bone quantity surrogate and is well established in patient care.
96
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Both clinical practice and experimental research still lack a noninvasive and precise method for evaluating fracture healing. Highprecision imaging instruments have been combined with refined mathematical structure models (finite element analysis) and superior computational power in investigations of several individual and combined scan parameters, and have been applied in the evaluation of fracture and graft healing with significant and promising results (Aro et al., 1989; Markel et al., 1991; Reynolds et al., 2007; Shefelbine et al., 2005). Compared to intact bone, in which geometrical measurements such as bone size, cross-sectional area and area moment of inertia have been shown to predict up to 70–80% of strength (Augat and Schorlemmer, 2006), fracture mineralization is more complicated due to both the fracture configuration and the overlapping bone repair processes, including callus formation, resorption and bone remodelling (Powell et al., 1989). We know that secondary bone healing (the most common form of diaphyseal healing seen in the clinic) is characterized by dynamic bone tissue that exhibits a simultaneously increasing degree of mineralization and callus resorption. There is thus a theoretical potential to utilize QCT as a tool to monitor fracture healing by segmenting QCT images into bone tissue with different degrees of mineralization and studying them separately, correlating them with biomechanical parameters. To the best of our knowledge, no such correlations have been reported previously. On the basis of these considerations, we used a well-established experimental fracture model and a clinically applicable QCT analysis technique to further investigate the relationship between biomechanical properties and QCT in flexibly fixed diaphyseal tibial fractures. The null hypothesis of our study was that bone strength is not correlated with any QCT parameters. 2. Methods 2.1. Animals and surgical procedures Forty male Wistar rats (Møllegårds Avlslaboratorium, Eiby, Denmark) with a mean weight of 379 g were used in this study. The animals were housed two per cage in rodent cages with a lid with a hinged water bottle divider and separate food areas, and under a 12h/12-h light/dark cycle. The rats received a standard rodent diet [RM3 (E), Special Diets Services, Witham, United Kingdom]. The experiments conformed to the Norwegian Council of Animal Research Code for the Care and Use of Animals for Experimental Purposes. Anaesthesia was effected using 0.3 g/100 ml of a working solution of fentanyl/fluanisone (Hypnorm, Janssen Pharmaceutica, Beerse, Belgium) and midazolam (Dormicum, Roche, Basel, Switzerland) administered subcutaneously (Smith, 2001). The left tibia was exposed through a 20-mm anterior incision from the tibial tuberosity in the distal direction. The muscles on the medial and lateral aspects of the tibia were carefully elevated from the tibia and the anterior twothirds of the bone was cut at the level of the anterior ridge using a fine-toothed circular saw blade mounted on an electric drill. The remaining one-third was then manually broken, leaving the fibula intact. The rats were initially treated with either EF (n = 20) or IMN (n = 20). The process of EF has been described previously (Mark et al., 2003; Sigurdsen et al., 2009). Briefly, a hole of diameter 0.67 mm was drilled in the tibia. Four threaded steel pins (diameter = 1.0 mm) were inserted: two proximal and two distal to the fracture. The fixator offset was 6 mm. The external fixator was placed anterolaterally, providing freedom of knee and ankle movement and apparent normal quadrupedal locomotion. In the IMN group, a 0.8-mm-diameter nail was inserted from the proximal side into the bone-marrow cavity by introducing the nail through the anterior tip of the tibial plateau, proceeding past the distal tibiofibular junction with the knee in a flexed position, and cut flush to the bony surface at the insertion side.
The nail was not reamed or locked. The surrounding soft tissue was handled carefully, leaving the medial and posterior muscle segments attached to the bone. The operation wound was closed in two layers with absorbable suture. A layer of transparent film dressing was then sprayed onto the sutured wound as protection. The alignment and accurate reduction of the fracture was verified perioperatively both visually and manually. Analgesia was maintained by injecting buprenorphine (0.05 mg/kg; Temgesic, Reckitt & Benckiser, Slough, UK) subcutaneously twice daily for the first three post-operative days. The animals were observed by the surgeon daily for three postoperative days after both initial and conversion surgery, and in the remaining experimental period on a weekly basis. Half of the animals in each treatment group were killed by an intraperitoneal injection of pentobarbital (100 mg/kg; Mebumal, Nycomed, Roskilde, Denmark) at 60 days after surgery, as previous studies have shown that rat leg fractures regain the strength similar to that of intact bones, after 8 weeks (Ekeland et al., 1981). Since we required information from the early phase of healing, the other half of the two treatment groups was killed at 30 days post-surgery. The tibias were dissected and examined visually and the external devices and nails were carefully removed. Between dissection and radiological, densitometric and mechanical examination, the bones were kept frozen at − 80 °C. The nails and the external fixator system, except for the four pins, were removed before QCT and mechanical testing. 2.2. Bone evaluations X-ray imaging was performed using a standard clinical digital system (Axiom Aristos, Siemens, München, Germany) to confirm bone healing. The X-ray tube settings were 46 kV, 1.0 mAs and a focus-to-film (source-to-image-receptor) distance of 115 cm. A standard micro-CT (μCT) system (Micro CAT II, Imtek—now Siemens) with 300 steps and 200° of rotation was used. The X-ray camera detector size was 2048 × 2048 with a bin factor of 2. The exposure time was 500 ms and the voxels were cubes with a side length of 50.7 μm. The images were reconstructed from scan data from both a narrow region of interest (ROI; 1.25 mm, 25 slices) near the fracture site, and a wide ROI (3.75 mm, 75 slices) encompassing more of the fracture callus region (Fig. 1). The μCT software's beamhardening error correction was employed. The reconstructed images were analysed with a commercially available reconstruction and visualization software package (Amira v4.1, Mercury Computer Systems, Mérignac Cedex, France). Simultaneous scanning of a Lucite phantom was performed with hydroxyapatite (HA) densities equal to 50, 250 and 750 mg/cm3, and Hounsfield units were linearly converted into HA densities (Supplement 1). Since QCT is predominantly applied to the densitometric examination of intact bones, there is no clearly defined consensus on density thresholds between fracture repair tissue such as soft and hard calluses and remodelled cortical bone (Augat et al., 1998; Claes et al., 2009; Korkusuz et al., 2000; Muller and Ruegsegger, 1997; Ward et al., 2005). We segmented the voxels into three categories based on voxel HA densities: soft callus (171–540 mg/cm3), hard callus (540– 1200 mg/cm3) and cortical bone (N1200 mg/cm3). The volume of each bone tissue category was calculated. Selection of these threshold values was based upon previous experiments and careful visual examination of the CT images with standard bone windows and levels (Lang et al., 1997; Reich et al., 1976). vBMD was calculated as the mean density of all voxels in the ROI. The tibias were ultimately placed in between gauze pads soaked with 0.9% saline before a bending test was performed using a universal testing machine with a servohydraulic mechanical linear drive actuator with 100 mm of total vertical displacement and a maximum axial tension loading capacity of 250 N (MTS 858 Mini Bionix, MTS Systems, Eden Prairie, MN). The set-up included a cantilever bending test that was designed to test the diaphyseal
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
97
adjusted for surgical treatment (EF or IMN) and time after surgery (30 or 60 days) to obtain R2 values. 3. Results Radiographically, all fractures demonstrated signs of healing, with gradual bridging of the fracture lines. From 30 to 60 days, QCTmeasured callus volumes and vBMD in the wide ROI (Table 1) were significantly decreased and cortical-bone volume was significantly increased. A resembling pattern was found for the narrow ROI (Table 2). A one-way ANOVA revealed no significant differences in mean cortical-bone volume between intact tibias and the 60-day treatment groups. The correlation data between strength and the QCT parameters are presented in Tables 3 (wide ROI) and 4 (narrow ROI). In the IMN group both cortical-bone volume and vBMD in the narrow ROI were correlated significantly and positively with strength in the early phase of healing; scatter plots of strength versus cortical-bone volume and vBMD are presented in Fig. 2. 4. Discussion
Fig. 1. Three-dimensional computed tomography (CT) reconstruction of a rat tibia at 60 days after initial treatment with intramedullary nailing (IMN), showing the fracture line (A), and narrow region of interest (ROI) (B), and the wide ROI (C). The wide ROI is also shown enlarged. The fibula has been resected.
fracture site, and a standard program that set the vertical travel speed to 160 mm/min, as described previously (Engesaeter et al., 1978). The load–displacement curve yielded the basic mechanical structural properties, maximum bending load, stiffness and energy to fracture using a commercial mathematical software package (Origin v 7.5, OriginLab, Northampton, MA) (Panjabi and White, 2001). A similar mechanical test performed on cadaveric tibia-implant constructions for EF (n = 5) and IMN (n = 5) yielded a mean (SEM) initial mechanical bending stiffness of 3.2 (0.6) and 0.6 (0.1) N/mm, respectively. In comparison, the stiffness of the intact tibias (n = 5) was 3.6 (0.3) N/mm and the maximum diaphyseal bending load was 27 (1.1) N.
2.3. Statistics Statistical data are expressed as mean and SEM values. Based on the biomechanical results reported by Utvag et al. (2003), we performed a power analysis to assess the number of animals needed for adequate statistical power. With a one-tailed test, a level of statistical significance at p b 0.05 and a strength of 0.80, the calculated number of animals needed in each group to detect an anticipated significant mean difference was eight. We therefore chose to include ten animals in each of the three groups. We used a two-way ANOVA to test for significant differences in means between the four study groups. A t-test was performed, and t-test probabilities were interpreted with the Bonferroni correction when applicable. We calculated Pearson's correlation coefficient for all four study groups. Significant correlations were explored with a linear regression analysis, with strength as the dependent variable and QCT parameters (soft-callus volume, hard-callus volume, cortical-bone volume and vBMD) as the independent variables. Linear regression analysis was also performed for the study groups pooled (n = 40) and then
As expected, QCT revealed a quantifiable characteristic pattern of secondary fracture healing in both treatment groups, with pronounced callus formation in the early phase and callus resorption with bone remodelling in the late phase. Bone strength showed a significant and positive correlation with QCT-measured fracture-site cortical-bone volume and vBMD in the IMN group in the early phase. This paper presents an analysis of statistical correlations between ultimate bending load and QCT-measured volumes of bone tissue with various degrees of mineralization and vBMD. Stiffness is another frequently used surrogate marker for maturation of fracture healing (Richardson et al., 1994). From a clinical point of view, the most important biomechanical property is the maximum (bending and torsional) load, which corresponds to the ability of the patient's leg to resist high load before a fracture or irreversible deformation occurs. In theory, a torsional test evaluates the weakest region of the bone, whereas the cantilever test can be designed to test a specific site (e.g. the fracture site) (Engesaeter et al., 1978). It should be noted that no histological evaluation of the tibias was performed in this study, and our experimental model utilized only an open osteotomy/fracture with moderate soft-tissue damage, which is not directly comparable to the closed fractures or open high-energy fractures that are often seen in the clinic. Table 1 Quantitative computed tomography (QCT) data from a wide region of interest (ROI; 3.75 mm, 75 slices) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either external fixation (EF) or intramedullary nailing (IMN). The rightmost column contains scan data from a similar ROI in intact tibias (n = 5). Measured bone tissue volumes are segmented into soft callus (171– 540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on voxel hydroxyapatite (HA) densities. Volumetric bone mineral density (vBMD) was also calculated. Data are mean (SEM) values.
3
Soft callus (mm )
Hard callus (mm3)
Cortical bone (mm3)
vBMD (mg/cm3)
a
30 days 60 days pa 30 days 60 days pa 30 days 60 days pa 30 days 60 days pa
Differences in means (t-test).
EF
IMN
16.2 (2.9) 6.3 (0.8) 0.004 35.1 (4.4) 11.5 (1.3) b0.001 10.2 (1.6) 19.8 (1.0) b0.001 839 (45) 1086 (30) 0.001
15.8 (2.2) 9.8 (0.9) 0.035 39.0 (2.8) 23.8 (1.9) b0.001 9.4 (1.6) 18.8 (1.4) b0.001 818 (33) 974 (23) 0.001
Intact tibias 6.1 (0.6)
5.7 (0.4)
16.6 (0.6)
1107 (24)
98
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Table 2 QCT data from a narrow ROI (1.25 mm, 25 slices) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either EF or IMN. The rightmost column contains scan data from a similar ROI in intact tibias (n = 5). Measured bone tissue volumes are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on voxel HA densities. vBMD values were also calculated.
Soft callus (mm3)
Hard callus (mm3)
3
Cortical bone (mm )
vBMD (mg/cm3)
a
30 days 60 days pa 30 days 60 days pa 30 days 60 days pa 30 days 60 days pa
EF
IMN
3.3 (0.9) 1.6 (0.2) 0.008 13 (1.7) 4.3 (0.5) b0.001 2.4 (0.6) 6.5 (0.4) b0.001 789 (50) 861 (150) 0.654
3.0 (0.7) 2.4 (0.3) 0.034 15 (0.9) 10 (0.7) 0.001 1.9 (0.4) 5.1 (0.6) b0.001 762 (36) 935 (22) 0.001
Intact tibias 2.1 (0.3)
Table 4 Correlation coefficients (Pearson's) between mechanical bending strength and segmented QCT parameters from a narrow ROI (1.25 mm, 25 slice) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either EF or IMN. Data are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N 1200 mg/cm3) based on QCT values. vBMD values were also calculated. Significant correlation coefficients are presented with pvalues and R2-values from linear regression analysis. The rightmost column presents standardized coefficient beta (p) values from a linear regression analysis between strength and each QCT parameter, adjusted for treatment and time after surgery, and with all study groups pooled (n = 40).
1.9 (0.1) Soft-callus volume 5.5 (0.2)
Hard-callus volume Cortical-bone volume
1102 (27) vBMD a
Differences in means (t-test).
b
Both treatment groups displayed the formation of a massive callus in the early phase. We know that callus formation is increased with reduced stability of the bone–implant system (Chao et al., 1989). Bone repair involves numerous processes that take place more or less simultaneously, and can be divided into (1) the early, inflammatory response, (2) soft and hard-callus formation through endochondral and intramembranous ossification (cartilage formation, calcification and removal) and (3) osteon remodelling (Einhorn, 1998). Remodelling is dependent on the necessary stability provided (e.g. by fracture callus or fracture fixation devices). Fractures that are very rigidly fixed exhibit primary healing, or bone repair, with diminished or absence of callus formation. Rigid fracture fixation of tibial shaft fractures is rare in the clinic (Court-Brown, 2006). The measured densities of the soft and hard calluses were not significantly correlated with strength. When all study groups were pooled (n = 40), strength tended to be correlated negatively with soft callus volume. In a clinical setting, the visible periosteal callus on an Xray is not a direct sign of fracture strength. The presence of a callus denotes a normal course of healing in a flexibly fixed fracture after some time, and signals that remodelling is impending or perhaps already advancing, with consequent bone strengthening. Our results indicate that even though the callus provides stability for the remodelling process, its biomechanical splinting effect is not a major contributor to the maximum bending strength of the bone. In theory, this complicated mineralization with the considerable formation and resorption of dynamic callus tissue may be difficult to describe clearly with summed or averaged densitometric parameters such as BMD or vBMD. Table 3 Correlation coefficients (Pearson's) between mechanical bending strength and QCT parameters from a wide ROI (3.75 mm, 75 slice) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with EF or IMN. Data are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on QCT values. vBMD values were also calculated. The rightmost column presents standardized coefficient beta (p) values from a linear regression analysis between strength and each QCT parameter, adjusted for treatment and time after surgery, and for all study groups pooled (n = 40).
Soft-callus volume Hard-callus volume Cortical-bone volume vBMD
30 days 60 days 30 days 60 days 30 days 60 days 30 days 60 days
EF
IMN
n = 40
–0.39 –0.41 0.15 –0.33 0.02 0.56 0.10 0.62
–0.46 0.61 –0.45 0.48 0.52 0.21 0.60 –0.29
–0.227 (0.169) 0.060 (0.980) 0.368 (0.068) 0.272 (0.168)
30 days 60 days 30 days 60 days 30 days 60 days 30 days 60 days
EF
IMN
n = 40
–0.46 –0.31 0.48 –0.10 –0.41 0.39 0.05 0.50
–0.51 0.43 –0.35 0.12 0.67a 0.44 0.68b 0.10
–0.285 (0.076) 0.202 (0.352) 0.281 (0.203) 0.270 (0.051)
p = 0.024. p = 0.021.
Only the fracture-site QCT-measured cortical-bone volume in the IMN group at 30 days correlated positively and significantly with strength. Almost half the variability of strength can be accounted for by variation in fracture-site cortical-bone volume. At 60 days, the cortical-bone volume in both treatment groups was similar to that of intact bone, which suggests that cortical bridging had already occurred. Cortical-bone remodelling with formation of new, strong lamellar structures across the fracture gap is the main factor responsible for the restoration of mechanical strength. Our results support the idea that the proximity of the ROI of QCT to the fracture influences the biomechanical importance of the measurement. It may be that the QCT-measured cortical-bone volume is sensitive to biomechanically important mineralization processes such as cortical bridging or osteon remodelling in this situation when the ROI is close to the fracture line in the early phase of healing in relatively flexibly fixed fractures. The higher bone–implant stiffness in the EF group may be at least partly responsible for a smaller amount of hard callus forming, which in turn probably reduces the stability for the cortical remodelling process and bone maturation. Possible subsequent differences in lamellar structure and microstructural properties may explain the relatively weak correlation between strength and the measured cortical-bone volume. Variation in the fracture configuration may also explain the relatively weak relationship between strength and cortical-bone volume. The ROI of QCT was aligned with the saw osteotomy perpendicular to the tibia and not necessarily with the ‘remainingthird’ fracture; the manual manipulation after partial osteotomy to fracture the remaining third of the tibial bone may have led to a variety of fracture configurations. As illustrated by Fig. 3, the accumulation of measured cortical bone represented by the curve demonstrated a minimum in the centre of the ROI, which probably expresses the fracture line and advocates that our ROIs are centred around it. No QCT study group measurements in the wide ROI were correlated significantly with strength. With all groups pooled (n = 40), cortical bone in the wide ROI tended to be positively correlated with strength. The wide ROI includes a relative large proportion of unfractured bone and may therefore have yielded a weaker correlation to bone strength. The calculated tissue volumes are based on the measured CT value of each voxel. Even high-resolution CT systems have limitations in visualizing bone microstructure (Augat and Schorlemmer, 2006). In heterogeneous materials (e.g. fractured bone) the CT value corresponds to the average attenuation contributed by all materials and chemical elements within its boundaries, and this is commonly
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
99
Fig. 2. Scatter plots of bending strength versus volumetric bone mineral density (vBMD) and fracture-site cortical-bone volume measured by quantitative CT in rat tibias at 30 days after diaphyseal fracture and treatment with IMN.
referred to as the partial volume effect (Augat and Eckstein, 2008; Kalender, 2005). The CT value thus reflects local radiopacity and is an approximation of the true amount and distribution of mineral density of the scanned subject. Rats are used widely in experimental studies in general, and in fracture healing research in particular. Still, one must be cautious in extrapolating findings from animal studies to human medical practice (Liebschner, 2004). Even though the relationship between image resolution and specimen size in our set-up is comparable to clinical CT systems and human tibias, results from a high-resolution μCT analysis of small-animal bones cannot be directly extrapolated to and applied in the clinic. When internal fixation devices are present in the clinic, QCT imaging precision is affected. Interestingly, with our set-up we identified significant and positive correlations between strength and QCT parameters. These parameters hold a clinically interesting potential, for example used as a supplement in experimental and clinical fracture evaluation in certain situations. The relatively modest bone-strength-predicting ability of our QCT parameters may be improved by refinement of threshold selection, image resolution and ROI configuration. In conclusion, even though we were able to quantify the characteristic pattern of secondary healing with early callus formation and late callus resorption with bone remodelling using QCT, segmenting the QCT data into these different types of bone tissue
did not increase the correlation with strength relative to using vBMD. The two significant correlations found between strength and QCTmeasured cortical bone and vBMD suggest that fracture-site QCT measurements are sensitive to the biomechanically important fracture mineralization in the early phase of healing in fractures with relatively flexible fixation. Conflict of interest The authors have all participated in data collection, data analysis and the writing of this manuscript. Sigurdsen, Reikeras and Utvag performed the animal surgery. This study was funded by the University of Oslo, and Oslo University Hospital Rikshospitalet. All authors have no conflicts of interest. Acknowledgements Financial support for this study was provided by the University of Oslo, Faculty Division Akershus University Hospital and Institute of Surgical Research, Oslo University Hospital Rikshospitalet, Oslo, Norway. All the authors participated in the study design, data collection, and analysis and interpretation of the data, while Sigurdsen, Reikeras and Utvag performed the animal surgery. The authors wish to thank engineer Hong Qu, PhD, and radiologist Ragnhild Gunderson, MD, at the Department of Radiology, Oslo University Hospital, Rikshospitalet for practical guidance and valuable assistance in μCT scanning and reconstruction. References
Fig. 3. Slice-by-slice average voxel distribution analysis of the CT-measured bone tissue (soft callus, hard callus and cortical bone) of transverse CT slices along the longitudinal 3.75-mm-long ROI centred around the fracture site. Slice data are averaged for all fractures in all study groups (n = 40). Voxels were cubes with a side length of 50.7 μm.
Aro, H.T., Wippermann, B.W., Hodgson, S.F., Wahner, H.W., Lewallen, D.G., Chao, E.Y., 1989. Prediction of properties of fracture callus by measurement of mineral density using micro-bone densitometry. J. Bone Joint Surg. Am. 71, 1020–1030. Augat, P., Eckstein, F., 2008. Quantitative imaging of musculoskeletal tissue. Annu. Rev. Biomed. Eng. 10, 369–390. Augat, P., Schorlemmer, S., 2006. The role of cortical bone and its microstructure in bone strength. Age Ageing 35 (Suppl 2), ii27–ii31. Augat, P., Gordon, C.L., Lang, T.F., Iida, H., Genant, H.K., 1998. Accuracy of cortical and trabecular bone measurements with peripheral quantitative computed tomography (pQCT). Phys. Med. Biol. 43, 2873–2883. Chao, E.Y., Aro, H.T., Lewallen, D.G., Kelly, P.J., 1989. The effect of rigidity on fracture healing in external fixation. Clin. Orthop. Relat. Res. 241, 24–35. Claes, L., Blakytny, R., Gockelmann, M., Schoen, M., Ignatius, A., Willie, B., 2009. Early dynamization by reduced fixation stiffness does not improve fracture healing in a rat femoral osteotomy model. J. Orthop. Res. 27, 22–27. Court-Brown, C.M., 2006. Fractures of the tibia and fibula, In: Bucholz, R.W., Heckman, J.D., Court-Brown, C. (Eds.), Rockwood and Green's Fractures in Adults Volume 2, sixth ed. Lippincott Williams & Wilkins, Philadelphia, pp. 2079–2146. Einhorn, T.A., 1998. The cell and molecularbiology of fracture healing. Clin. Orthop. Relat. Res. 355S, S7–S21.
100
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Ekeland, A., Engesaeter, L.B., Langeland, N., 1981. Mechanical properties of fractured and intact rat femora evaluated by bending, torsional and tensile tests. Acta Orthop. Scand. 52, 605–613. Engesaeter, L.B., Ekeland, A., Langeland, N., 1978. Methods for testing the mechanical properties of the rat femur. Acta Orthop. Scand. 49, 512–518. Kalender, W.A., 2005. Computed Tomography, second ed. Publicis, Erlangen, Germany. Korkusuz, F., Akin, S., Akkus, O., Korkusuz, P., 2000. Assessment of mineral density and atomic content of fracture callus by quantitative computerized tomography. J. Orthop. Sci. 5, 248–255. Lang, T.F., Keyak, J.H., Heitz, M.W., Augat, P., Lu, Y., Mathur, A., et al., 1997. Volumetric quantitative computed tomography of the proximal femur: precision and relation to bone strength. Bone 21, 101–108. Liebschner, M.A., 2004. Biomechanical considerations of animal models used in tissue engineering of bone. Biomaterials 25, 1697–1714. Mark, H., Bergholm, J., Nilsson, A., Rydevik, B., Stromberg, L., 2003. An external fixation method and device to study fracture healing in rats. Acta Orthop. Scand. 74, 476–482. Markel, M.D., Wikenheiser, M.A., Morin, R.L., Lewallen, D.G., Chao, E.Y., 1990. Quantification of bone healing. Comparison of QCT, SPA, MRI, and DEXA in dog osteotomies. Acta Orthop. Scand. 61, 487–498. Markel, M.D., Morin, R.L., Wikenheiser, M.A., Lewallen, D.G., Chao, E.Y., 1991. Quantitative CT for the evaluation of bone healing. Calcif. Tissue Int. 49, 427–432. Muller, R., Ruegsegger, P., 1997. Micro-tomographic imaging for the nondestructive evaluation of trabecular bone architecture. Stud. Health Technol. Inform. 40, 61–79. Panjabi, M.M., White, A.A., 2001. Materials under loads, Biomehanics in the Musculoskeletal System, first ed. Churchill Livingstone, Philadelphia, pp. 65–127.
Powell, E.S., Lawford, P.V., Duckworth, T., Black, M.M., 1989. Is callus calcium content an indicator of the mechanical strength of healing fractures? An experimental study in rat metatarsals. J. Biomed. Eng. 11, 277–281. Reich, N.E., Seidelmann, F.E., Tubbs, R.R., MacIntyre, W.J., Meaney, T.F., Alfidi, R.J., et al., 1976. Determination of bone mineral content using CT scanning. Am. J. Roentgenol. 127, 593–594. Reynolds, D.G., Hock, C., Shaikh, S., Jacobson, J., Zhang, X., Rubery, P.T., et al., 2007. Micro-computed tomography prediction of biomechanical strength in murine structural bone grafts. J. Biomech. 40, 3178–3186. Richardson, J.B., Cunningham, J.L., Goodship, A.E., O'Connor, B.T., Kenwright, J., 1994. Measuring stiffness can define healing of tibial fractures. J. Bone Joint Surg. Br. 76, 389–394. Shefelbine, S.J., Simon, U., Claes, L., Gold, A., Gabet, Y., Bab, I., et al., 2005. Prediction of fracture callus mechanical properties using micro-CT images and voxel-based finite element analysis. Bone 36, 480–488. Sigurdsen, U.E., Reikeras, O., Utvag, S.E., 2009. External fixation compared to intramedullary nailing of tibial fractures in the rat. Acta Orthop. 80, 375–379. Smith, A., 2001. Laboratory Animal Science, first ed. Norwegian School of Veterinary Science, Oslo, Norway. Utvag, S.E., Grundnes, O., Rindal, D.B., Reikeras, O., 2003. Influence of extensive muscle injury on fracture healing in rat tibia. J. Orthop. Trauma 17, 430–435. Ward, K.A., Adams, J.E., Hangartner, T.N., 2005. Recommendations for thresholds for cortical bone geometry and density measurement by peripheral quantitative computed tomography. Calcif. Tissue Int. 77, 275–280.
Contents lists available at ScienceDirect
Clinical Biomechanics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c l i n b i o m e c h
Correlations between strength and quantitative computed tomography measurement of callus mineralization in experimental tibial fractures Ulf Sigurdsen a,b,⁎, Olav Reikeras c, Arne Hoiseth d, Stein Erik Utvag e,a a
Department of Orthopedic Surgery, University of Oslo, Norway Institute of Surgical Research, Oslo University Hospital Rikshospitalet, Norway Department of Orthopedic Surgery, Oslo University Hospital Rikshospitalet, Norway d Curato Roentgen Inc., Oslo, Norway e Department of Orthopedic Surgery, Akershus University Hospital, Norway b c
a r t i c l e
i n f o
Article history: Received 21 April 2010 Accepted 8 September 2010 Keywords: Experimental fracture healing Tibial diaphyseal fracture Quantitative computed tomography External fixation Intramedullary nailing
a b s t r a c t Background: The evaluation of fracture healing in the clinic has not changed significantly during the past few decades, despite the development of modern tissue-imaging tools. Recent publications have reported significant and interesting associations between biomechanical properties and quantitative computed tomography data of fractures and grafts. We therefore studied the correlations between the strength and segmented quantitative computed tomography data of tibial diaphyseal fractures. Methods: Forty male rats received a tibial-shaft osteotomy that was initially stabilized with either intramedullary nailing or external fixation. Evaluation at 30 and 60 days post-osteotomy included X-ray, quantitative computed tomography and bending testing. Quantitative computed tomography data were segmented by voxel density into soft callus (171–539 mg/cm3), hard callus (540–1199 mg/cm3) and cortical bone (≥ 1200 mg/cm3), and volumetric bone mineral density was calculated. Findings: All fractures demonstrated pronounced formation of soft and hard callus tissues at 30 days postosteotomy, and at 60 days the cortical bone volume was significantly increased with callus resorption. Bending strength correlated significantly and positively with fracture-site cortical bone volume and volumetric bone mineral density in the intramedullary nailed group in the early phase of healing. Interpretation: Quantitative computed tomography was used to quantify characteristic secondary healing. The observed correlations indicate that biomechanically important mineralization can be measured by quantitative computed tomography in the early phase of healing in flexibly fixed fractures. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Over the past few decades, fracture healing has been evaluated by patient interview, clinical examination and standard biplane radiograms. The dramatic developments in advanced tissue-imaging techniques and recent advances within computational science have not yet provided the orthopaedic surgeon with non-invasive methods that are more accurate at evaluating bone repair. Several authors have attributed this to the normally good fracture prognosis. However, an increasing number of patients present with co-morbidities that impair bone quality and repair, such as age-related osteoporosis and diabetes. There is also an increased demand for accurate staging of fracture healing for the early and safe removal of restrictions and surgical implants. Tibial
⁎ Corresponding author. Institute of Surgical Research, Oslo University Hospital Rikshospitalet, 0027 Oslo, Norway. E-mail address: [email protected] (U. Sigurdsen). 0268-0033/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2010.09.004
diaphyseal fractures can be among the most difficult fractures to treat, and the surgeon's choice is often between intramedullary nailing (IMN) and external fixation (EF) (Court-Brown, 2006). Measurement of the mineralization of intact bone provides a valuable estimation of bone quantity and quality, and future fracture risk (Markel et al., 1990). We know that exceptional biomechanical properties such as toughness characterize bone tissue, and that these properties are associated mainly with the amount and histological organization of calcium minerals. Other biomechanically relevant, but less-measurable tissue factors include bone microstructure, mineral crystal size, and the number and size of bone pores (Powell et al., 1989). Intact bone mineralization measured by several techniques is statistically correlated with biomechanical properties and clinically valuable fracture risk predictors (Markel et al., 1990). Measurement of bone mass—either as bone mineral density (BMD; area density in grams per square centimetre) or bone mineral content (in grams)—by dual X-ray absorptiometry is one such method for quantitative bone assessment. BMD combined with standard age–gender tables is a preferred bone quantity surrogate and is well established in patient care.
96
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Both clinical practice and experimental research still lack a noninvasive and precise method for evaluating fracture healing. Highprecision imaging instruments have been combined with refined mathematical structure models (finite element analysis) and superior computational power in investigations of several individual and combined scan parameters, and have been applied in the evaluation of fracture and graft healing with significant and promising results (Aro et al., 1989; Markel et al., 1991; Reynolds et al., 2007; Shefelbine et al., 2005). Compared to intact bone, in which geometrical measurements such as bone size, cross-sectional area and area moment of inertia have been shown to predict up to 70–80% of strength (Augat and Schorlemmer, 2006), fracture mineralization is more complicated due to both the fracture configuration and the overlapping bone repair processes, including callus formation, resorption and bone remodelling (Powell et al., 1989). We know that secondary bone healing (the most common form of diaphyseal healing seen in the clinic) is characterized by dynamic bone tissue that exhibits a simultaneously increasing degree of mineralization and callus resorption. There is thus a theoretical potential to utilize QCT as a tool to monitor fracture healing by segmenting QCT images into bone tissue with different degrees of mineralization and studying them separately, correlating them with biomechanical parameters. To the best of our knowledge, no such correlations have been reported previously. On the basis of these considerations, we used a well-established experimental fracture model and a clinically applicable QCT analysis technique to further investigate the relationship between biomechanical properties and QCT in flexibly fixed diaphyseal tibial fractures. The null hypothesis of our study was that bone strength is not correlated with any QCT parameters. 2. Methods 2.1. Animals and surgical procedures Forty male Wistar rats (Møllegårds Avlslaboratorium, Eiby, Denmark) with a mean weight of 379 g were used in this study. The animals were housed two per cage in rodent cages with a lid with a hinged water bottle divider and separate food areas, and under a 12h/12-h light/dark cycle. The rats received a standard rodent diet [RM3 (E), Special Diets Services, Witham, United Kingdom]. The experiments conformed to the Norwegian Council of Animal Research Code for the Care and Use of Animals for Experimental Purposes. Anaesthesia was effected using 0.3 g/100 ml of a working solution of fentanyl/fluanisone (Hypnorm, Janssen Pharmaceutica, Beerse, Belgium) and midazolam (Dormicum, Roche, Basel, Switzerland) administered subcutaneously (Smith, 2001). The left tibia was exposed through a 20-mm anterior incision from the tibial tuberosity in the distal direction. The muscles on the medial and lateral aspects of the tibia were carefully elevated from the tibia and the anterior twothirds of the bone was cut at the level of the anterior ridge using a fine-toothed circular saw blade mounted on an electric drill. The remaining one-third was then manually broken, leaving the fibula intact. The rats were initially treated with either EF (n = 20) or IMN (n = 20). The process of EF has been described previously (Mark et al., 2003; Sigurdsen et al., 2009). Briefly, a hole of diameter 0.67 mm was drilled in the tibia. Four threaded steel pins (diameter = 1.0 mm) were inserted: two proximal and two distal to the fracture. The fixator offset was 6 mm. The external fixator was placed anterolaterally, providing freedom of knee and ankle movement and apparent normal quadrupedal locomotion. In the IMN group, a 0.8-mm-diameter nail was inserted from the proximal side into the bone-marrow cavity by introducing the nail through the anterior tip of the tibial plateau, proceeding past the distal tibiofibular junction with the knee in a flexed position, and cut flush to the bony surface at the insertion side.
The nail was not reamed or locked. The surrounding soft tissue was handled carefully, leaving the medial and posterior muscle segments attached to the bone. The operation wound was closed in two layers with absorbable suture. A layer of transparent film dressing was then sprayed onto the sutured wound as protection. The alignment and accurate reduction of the fracture was verified perioperatively both visually and manually. Analgesia was maintained by injecting buprenorphine (0.05 mg/kg; Temgesic, Reckitt & Benckiser, Slough, UK) subcutaneously twice daily for the first three post-operative days. The animals were observed by the surgeon daily for three postoperative days after both initial and conversion surgery, and in the remaining experimental period on a weekly basis. Half of the animals in each treatment group were killed by an intraperitoneal injection of pentobarbital (100 mg/kg; Mebumal, Nycomed, Roskilde, Denmark) at 60 days after surgery, as previous studies have shown that rat leg fractures regain the strength similar to that of intact bones, after 8 weeks (Ekeland et al., 1981). Since we required information from the early phase of healing, the other half of the two treatment groups was killed at 30 days post-surgery. The tibias were dissected and examined visually and the external devices and nails were carefully removed. Between dissection and radiological, densitometric and mechanical examination, the bones were kept frozen at − 80 °C. The nails and the external fixator system, except for the four pins, were removed before QCT and mechanical testing. 2.2. Bone evaluations X-ray imaging was performed using a standard clinical digital system (Axiom Aristos, Siemens, München, Germany) to confirm bone healing. The X-ray tube settings were 46 kV, 1.0 mAs and a focus-to-film (source-to-image-receptor) distance of 115 cm. A standard micro-CT (μCT) system (Micro CAT II, Imtek—now Siemens) with 300 steps and 200° of rotation was used. The X-ray camera detector size was 2048 × 2048 with a bin factor of 2. The exposure time was 500 ms and the voxels were cubes with a side length of 50.7 μm. The images were reconstructed from scan data from both a narrow region of interest (ROI; 1.25 mm, 25 slices) near the fracture site, and a wide ROI (3.75 mm, 75 slices) encompassing more of the fracture callus region (Fig. 1). The μCT software's beamhardening error correction was employed. The reconstructed images were analysed with a commercially available reconstruction and visualization software package (Amira v4.1, Mercury Computer Systems, Mérignac Cedex, France). Simultaneous scanning of a Lucite phantom was performed with hydroxyapatite (HA) densities equal to 50, 250 and 750 mg/cm3, and Hounsfield units were linearly converted into HA densities (Supplement 1). Since QCT is predominantly applied to the densitometric examination of intact bones, there is no clearly defined consensus on density thresholds between fracture repair tissue such as soft and hard calluses and remodelled cortical bone (Augat et al., 1998; Claes et al., 2009; Korkusuz et al., 2000; Muller and Ruegsegger, 1997; Ward et al., 2005). We segmented the voxels into three categories based on voxel HA densities: soft callus (171–540 mg/cm3), hard callus (540– 1200 mg/cm3) and cortical bone (N1200 mg/cm3). The volume of each bone tissue category was calculated. Selection of these threshold values was based upon previous experiments and careful visual examination of the CT images with standard bone windows and levels (Lang et al., 1997; Reich et al., 1976). vBMD was calculated as the mean density of all voxels in the ROI. The tibias were ultimately placed in between gauze pads soaked with 0.9% saline before a bending test was performed using a universal testing machine with a servohydraulic mechanical linear drive actuator with 100 mm of total vertical displacement and a maximum axial tension loading capacity of 250 N (MTS 858 Mini Bionix, MTS Systems, Eden Prairie, MN). The set-up included a cantilever bending test that was designed to test the diaphyseal
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
97
adjusted for surgical treatment (EF or IMN) and time after surgery (30 or 60 days) to obtain R2 values. 3. Results Radiographically, all fractures demonstrated signs of healing, with gradual bridging of the fracture lines. From 30 to 60 days, QCTmeasured callus volumes and vBMD in the wide ROI (Table 1) were significantly decreased and cortical-bone volume was significantly increased. A resembling pattern was found for the narrow ROI (Table 2). A one-way ANOVA revealed no significant differences in mean cortical-bone volume between intact tibias and the 60-day treatment groups. The correlation data between strength and the QCT parameters are presented in Tables 3 (wide ROI) and 4 (narrow ROI). In the IMN group both cortical-bone volume and vBMD in the narrow ROI were correlated significantly and positively with strength in the early phase of healing; scatter plots of strength versus cortical-bone volume and vBMD are presented in Fig. 2. 4. Discussion
Fig. 1. Three-dimensional computed tomography (CT) reconstruction of a rat tibia at 60 days after initial treatment with intramedullary nailing (IMN), showing the fracture line (A), and narrow region of interest (ROI) (B), and the wide ROI (C). The wide ROI is also shown enlarged. The fibula has been resected.
fracture site, and a standard program that set the vertical travel speed to 160 mm/min, as described previously (Engesaeter et al., 1978). The load–displacement curve yielded the basic mechanical structural properties, maximum bending load, stiffness and energy to fracture using a commercial mathematical software package (Origin v 7.5, OriginLab, Northampton, MA) (Panjabi and White, 2001). A similar mechanical test performed on cadaveric tibia-implant constructions for EF (n = 5) and IMN (n = 5) yielded a mean (SEM) initial mechanical bending stiffness of 3.2 (0.6) and 0.6 (0.1) N/mm, respectively. In comparison, the stiffness of the intact tibias (n = 5) was 3.6 (0.3) N/mm and the maximum diaphyseal bending load was 27 (1.1) N.
2.3. Statistics Statistical data are expressed as mean and SEM values. Based on the biomechanical results reported by Utvag et al. (2003), we performed a power analysis to assess the number of animals needed for adequate statistical power. With a one-tailed test, a level of statistical significance at p b 0.05 and a strength of 0.80, the calculated number of animals needed in each group to detect an anticipated significant mean difference was eight. We therefore chose to include ten animals in each of the three groups. We used a two-way ANOVA to test for significant differences in means between the four study groups. A t-test was performed, and t-test probabilities were interpreted with the Bonferroni correction when applicable. We calculated Pearson's correlation coefficient for all four study groups. Significant correlations were explored with a linear regression analysis, with strength as the dependent variable and QCT parameters (soft-callus volume, hard-callus volume, cortical-bone volume and vBMD) as the independent variables. Linear regression analysis was also performed for the study groups pooled (n = 40) and then
As expected, QCT revealed a quantifiable characteristic pattern of secondary fracture healing in both treatment groups, with pronounced callus formation in the early phase and callus resorption with bone remodelling in the late phase. Bone strength showed a significant and positive correlation with QCT-measured fracture-site cortical-bone volume and vBMD in the IMN group in the early phase. This paper presents an analysis of statistical correlations between ultimate bending load and QCT-measured volumes of bone tissue with various degrees of mineralization and vBMD. Stiffness is another frequently used surrogate marker for maturation of fracture healing (Richardson et al., 1994). From a clinical point of view, the most important biomechanical property is the maximum (bending and torsional) load, which corresponds to the ability of the patient's leg to resist high load before a fracture or irreversible deformation occurs. In theory, a torsional test evaluates the weakest region of the bone, whereas the cantilever test can be designed to test a specific site (e.g. the fracture site) (Engesaeter et al., 1978). It should be noted that no histological evaluation of the tibias was performed in this study, and our experimental model utilized only an open osteotomy/fracture with moderate soft-tissue damage, which is not directly comparable to the closed fractures or open high-energy fractures that are often seen in the clinic. Table 1 Quantitative computed tomography (QCT) data from a wide region of interest (ROI; 3.75 mm, 75 slices) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either external fixation (EF) or intramedullary nailing (IMN). The rightmost column contains scan data from a similar ROI in intact tibias (n = 5). Measured bone tissue volumes are segmented into soft callus (171– 540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on voxel hydroxyapatite (HA) densities. Volumetric bone mineral density (vBMD) was also calculated. Data are mean (SEM) values.
3
Soft callus (mm )
Hard callus (mm3)
Cortical bone (mm3)
vBMD (mg/cm3)
a
30 days 60 days pa 30 days 60 days pa 30 days 60 days pa 30 days 60 days pa
Differences in means (t-test).
EF
IMN
16.2 (2.9) 6.3 (0.8) 0.004 35.1 (4.4) 11.5 (1.3) b0.001 10.2 (1.6) 19.8 (1.0) b0.001 839 (45) 1086 (30) 0.001
15.8 (2.2) 9.8 (0.9) 0.035 39.0 (2.8) 23.8 (1.9) b0.001 9.4 (1.6) 18.8 (1.4) b0.001 818 (33) 974 (23) 0.001
Intact tibias 6.1 (0.6)
5.7 (0.4)
16.6 (0.6)
1107 (24)
98
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Table 2 QCT data from a narrow ROI (1.25 mm, 25 slices) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either EF or IMN. The rightmost column contains scan data from a similar ROI in intact tibias (n = 5). Measured bone tissue volumes are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on voxel HA densities. vBMD values were also calculated.
Soft callus (mm3)
Hard callus (mm3)
3
Cortical bone (mm )
vBMD (mg/cm3)
a
30 days 60 days pa 30 days 60 days pa 30 days 60 days pa 30 days 60 days pa
EF
IMN
3.3 (0.9) 1.6 (0.2) 0.008 13 (1.7) 4.3 (0.5) b0.001 2.4 (0.6) 6.5 (0.4) b0.001 789 (50) 861 (150) 0.654
3.0 (0.7) 2.4 (0.3) 0.034 15 (0.9) 10 (0.7) 0.001 1.9 (0.4) 5.1 (0.6) b0.001 762 (36) 935 (22) 0.001
Intact tibias 2.1 (0.3)
Table 4 Correlation coefficients (Pearson's) between mechanical bending strength and segmented QCT parameters from a narrow ROI (1.25 mm, 25 slice) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with either EF or IMN. Data are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N 1200 mg/cm3) based on QCT values. vBMD values were also calculated. Significant correlation coefficients are presented with pvalues and R2-values from linear regression analysis. The rightmost column presents standardized coefficient beta (p) values from a linear regression analysis between strength and each QCT parameter, adjusted for treatment and time after surgery, and with all study groups pooled (n = 40).
1.9 (0.1) Soft-callus volume 5.5 (0.2)
Hard-callus volume Cortical-bone volume
1102 (27) vBMD a
Differences in means (t-test).
b
Both treatment groups displayed the formation of a massive callus in the early phase. We know that callus formation is increased with reduced stability of the bone–implant system (Chao et al., 1989). Bone repair involves numerous processes that take place more or less simultaneously, and can be divided into (1) the early, inflammatory response, (2) soft and hard-callus formation through endochondral and intramembranous ossification (cartilage formation, calcification and removal) and (3) osteon remodelling (Einhorn, 1998). Remodelling is dependent on the necessary stability provided (e.g. by fracture callus or fracture fixation devices). Fractures that are very rigidly fixed exhibit primary healing, or bone repair, with diminished or absence of callus formation. Rigid fracture fixation of tibial shaft fractures is rare in the clinic (Court-Brown, 2006). The measured densities of the soft and hard calluses were not significantly correlated with strength. When all study groups were pooled (n = 40), strength tended to be correlated negatively with soft callus volume. In a clinical setting, the visible periosteal callus on an Xray is not a direct sign of fracture strength. The presence of a callus denotes a normal course of healing in a flexibly fixed fracture after some time, and signals that remodelling is impending or perhaps already advancing, with consequent bone strengthening. Our results indicate that even though the callus provides stability for the remodelling process, its biomechanical splinting effect is not a major contributor to the maximum bending strength of the bone. In theory, this complicated mineralization with the considerable formation and resorption of dynamic callus tissue may be difficult to describe clearly with summed or averaged densitometric parameters such as BMD or vBMD. Table 3 Correlation coefficients (Pearson's) between mechanical bending strength and QCT parameters from a wide ROI (3.75 mm, 75 slice) at the fracture site in tibial diaphyseal fractures in rats at 30 and 60 days following treatment with EF or IMN. Data are segmented into soft callus (171–540 mg/cm3), hard callus (540–1200 mg/cm3) and cortical bone (N1200 mg/cm3) based on QCT values. vBMD values were also calculated. The rightmost column presents standardized coefficient beta (p) values from a linear regression analysis between strength and each QCT parameter, adjusted for treatment and time after surgery, and for all study groups pooled (n = 40).
Soft-callus volume Hard-callus volume Cortical-bone volume vBMD
30 days 60 days 30 days 60 days 30 days 60 days 30 days 60 days
EF
IMN
n = 40
–0.39 –0.41 0.15 –0.33 0.02 0.56 0.10 0.62
–0.46 0.61 –0.45 0.48 0.52 0.21 0.60 –0.29
–0.227 (0.169) 0.060 (0.980) 0.368 (0.068) 0.272 (0.168)
30 days 60 days 30 days 60 days 30 days 60 days 30 days 60 days
EF
IMN
n = 40
–0.46 –0.31 0.48 –0.10 –0.41 0.39 0.05 0.50
–0.51 0.43 –0.35 0.12 0.67a 0.44 0.68b 0.10
–0.285 (0.076) 0.202 (0.352) 0.281 (0.203) 0.270 (0.051)
p = 0.024. p = 0.021.
Only the fracture-site QCT-measured cortical-bone volume in the IMN group at 30 days correlated positively and significantly with strength. Almost half the variability of strength can be accounted for by variation in fracture-site cortical-bone volume. At 60 days, the cortical-bone volume in both treatment groups was similar to that of intact bone, which suggests that cortical bridging had already occurred. Cortical-bone remodelling with formation of new, strong lamellar structures across the fracture gap is the main factor responsible for the restoration of mechanical strength. Our results support the idea that the proximity of the ROI of QCT to the fracture influences the biomechanical importance of the measurement. It may be that the QCT-measured cortical-bone volume is sensitive to biomechanically important mineralization processes such as cortical bridging or osteon remodelling in this situation when the ROI is close to the fracture line in the early phase of healing in relatively flexibly fixed fractures. The higher bone–implant stiffness in the EF group may be at least partly responsible for a smaller amount of hard callus forming, which in turn probably reduces the stability for the cortical remodelling process and bone maturation. Possible subsequent differences in lamellar structure and microstructural properties may explain the relatively weak correlation between strength and the measured cortical-bone volume. Variation in the fracture configuration may also explain the relatively weak relationship between strength and cortical-bone volume. The ROI of QCT was aligned with the saw osteotomy perpendicular to the tibia and not necessarily with the ‘remainingthird’ fracture; the manual manipulation after partial osteotomy to fracture the remaining third of the tibial bone may have led to a variety of fracture configurations. As illustrated by Fig. 3, the accumulation of measured cortical bone represented by the curve demonstrated a minimum in the centre of the ROI, which probably expresses the fracture line and advocates that our ROIs are centred around it. No QCT study group measurements in the wide ROI were correlated significantly with strength. With all groups pooled (n = 40), cortical bone in the wide ROI tended to be positively correlated with strength. The wide ROI includes a relative large proportion of unfractured bone and may therefore have yielded a weaker correlation to bone strength. The calculated tissue volumes are based on the measured CT value of each voxel. Even high-resolution CT systems have limitations in visualizing bone microstructure (Augat and Schorlemmer, 2006). In heterogeneous materials (e.g. fractured bone) the CT value corresponds to the average attenuation contributed by all materials and chemical elements within its boundaries, and this is commonly
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
99
Fig. 2. Scatter plots of bending strength versus volumetric bone mineral density (vBMD) and fracture-site cortical-bone volume measured by quantitative CT in rat tibias at 30 days after diaphyseal fracture and treatment with IMN.
referred to as the partial volume effect (Augat and Eckstein, 2008; Kalender, 2005). The CT value thus reflects local radiopacity and is an approximation of the true amount and distribution of mineral density of the scanned subject. Rats are used widely in experimental studies in general, and in fracture healing research in particular. Still, one must be cautious in extrapolating findings from animal studies to human medical practice (Liebschner, 2004). Even though the relationship between image resolution and specimen size in our set-up is comparable to clinical CT systems and human tibias, results from a high-resolution μCT analysis of small-animal bones cannot be directly extrapolated to and applied in the clinic. When internal fixation devices are present in the clinic, QCT imaging precision is affected. Interestingly, with our set-up we identified significant and positive correlations between strength and QCT parameters. These parameters hold a clinically interesting potential, for example used as a supplement in experimental and clinical fracture evaluation in certain situations. The relatively modest bone-strength-predicting ability of our QCT parameters may be improved by refinement of threshold selection, image resolution and ROI configuration. In conclusion, even though we were able to quantify the characteristic pattern of secondary healing with early callus formation and late callus resorption with bone remodelling using QCT, segmenting the QCT data into these different types of bone tissue
did not increase the correlation with strength relative to using vBMD. The two significant correlations found between strength and QCTmeasured cortical bone and vBMD suggest that fracture-site QCT measurements are sensitive to the biomechanically important fracture mineralization in the early phase of healing in fractures with relatively flexible fixation. Conflict of interest The authors have all participated in data collection, data analysis and the writing of this manuscript. Sigurdsen, Reikeras and Utvag performed the animal surgery. This study was funded by the University of Oslo, and Oslo University Hospital Rikshospitalet. All authors have no conflicts of interest. Acknowledgements Financial support for this study was provided by the University of Oslo, Faculty Division Akershus University Hospital and Institute of Surgical Research, Oslo University Hospital Rikshospitalet, Oslo, Norway. All the authors participated in the study design, data collection, and analysis and interpretation of the data, while Sigurdsen, Reikeras and Utvag performed the animal surgery. The authors wish to thank engineer Hong Qu, PhD, and radiologist Ragnhild Gunderson, MD, at the Department of Radiology, Oslo University Hospital, Rikshospitalet for practical guidance and valuable assistance in μCT scanning and reconstruction. References
Fig. 3. Slice-by-slice average voxel distribution analysis of the CT-measured bone tissue (soft callus, hard callus and cortical bone) of transverse CT slices along the longitudinal 3.75-mm-long ROI centred around the fracture site. Slice data are averaged for all fractures in all study groups (n = 40). Voxels were cubes with a side length of 50.7 μm.
Aro, H.T., Wippermann, B.W., Hodgson, S.F., Wahner, H.W., Lewallen, D.G., Chao, E.Y., 1989. Prediction of properties of fracture callus by measurement of mineral density using micro-bone densitometry. J. Bone Joint Surg. Am. 71, 1020–1030. Augat, P., Eckstein, F., 2008. Quantitative imaging of musculoskeletal tissue. Annu. Rev. Biomed. Eng. 10, 369–390. Augat, P., Schorlemmer, S., 2006. The role of cortical bone and its microstructure in bone strength. Age Ageing 35 (Suppl 2), ii27–ii31. Augat, P., Gordon, C.L., Lang, T.F., Iida, H., Genant, H.K., 1998. Accuracy of cortical and trabecular bone measurements with peripheral quantitative computed tomography (pQCT). Phys. Med. Biol. 43, 2873–2883. Chao, E.Y., Aro, H.T., Lewallen, D.G., Kelly, P.J., 1989. The effect of rigidity on fracture healing in external fixation. Clin. Orthop. Relat. Res. 241, 24–35. Claes, L., Blakytny, R., Gockelmann, M., Schoen, M., Ignatius, A., Willie, B., 2009. Early dynamization by reduced fixation stiffness does not improve fracture healing in a rat femoral osteotomy model. J. Orthop. Res. 27, 22–27. Court-Brown, C.M., 2006. Fractures of the tibia and fibula, In: Bucholz, R.W., Heckman, J.D., Court-Brown, C. (Eds.), Rockwood and Green's Fractures in Adults Volume 2, sixth ed. Lippincott Williams & Wilkins, Philadelphia, pp. 2079–2146. Einhorn, T.A., 1998. The cell and molecularbiology of fracture healing. Clin. Orthop. Relat. Res. 355S, S7–S21.
100
U. Sigurdsen et al. / Clinical Biomechanics 26 (2011) 95–100
Ekeland, A., Engesaeter, L.B., Langeland, N., 1981. Mechanical properties of fractured and intact rat femora evaluated by bending, torsional and tensile tests. Acta Orthop. Scand. 52, 605–613. Engesaeter, L.B., Ekeland, A., Langeland, N., 1978. Methods for testing the mechanical properties of the rat femur. Acta Orthop. Scand. 49, 512–518. Kalender, W.A., 2005. Computed Tomography, second ed. Publicis, Erlangen, Germany. Korkusuz, F., Akin, S., Akkus, O., Korkusuz, P., 2000. Assessment of mineral density and atomic content of fracture callus by quantitative computerized tomography. J. Orthop. Sci. 5, 248–255. Lang, T.F., Keyak, J.H., Heitz, M.W., Augat, P., Lu, Y., Mathur, A., et al., 1997. Volumetric quantitative computed tomography of the proximal femur: precision and relation to bone strength. Bone 21, 101–108. Liebschner, M.A., 2004. Biomechanical considerations of animal models used in tissue engineering of bone. Biomaterials 25, 1697–1714. Mark, H., Bergholm, J., Nilsson, A., Rydevik, B., Stromberg, L., 2003. An external fixation method and device to study fracture healing in rats. Acta Orthop. Scand. 74, 476–482. Markel, M.D., Wikenheiser, M.A., Morin, R.L., Lewallen, D.G., Chao, E.Y., 1990. Quantification of bone healing. Comparison of QCT, SPA, MRI, and DEXA in dog osteotomies. Acta Orthop. Scand. 61, 487–498. Markel, M.D., Morin, R.L., Wikenheiser, M.A., Lewallen, D.G., Chao, E.Y., 1991. Quantitative CT for the evaluation of bone healing. Calcif. Tissue Int. 49, 427–432. Muller, R., Ruegsegger, P., 1997. Micro-tomographic imaging for the nondestructive evaluation of trabecular bone architecture. Stud. Health Technol. Inform. 40, 61–79. Panjabi, M.M., White, A.A., 2001. Materials under loads, Biomehanics in the Musculoskeletal System, first ed. Churchill Livingstone, Philadelphia, pp. 65–127.
Powell, E.S., Lawford, P.V., Duckworth, T., Black, M.M., 1989. Is callus calcium content an indicator of the mechanical strength of healing fractures? An experimental study in rat metatarsals. J. Biomed. Eng. 11, 277–281. Reich, N.E., Seidelmann, F.E., Tubbs, R.R., MacIntyre, W.J., Meaney, T.F., Alfidi, R.J., et al., 1976. Determination of bone mineral content using CT scanning. Am. J. Roentgenol. 127, 593–594. Reynolds, D.G., Hock, C., Shaikh, S., Jacobson, J., Zhang, X., Rubery, P.T., et al., 2007. Micro-computed tomography prediction of biomechanical strength in murine structural bone grafts. J. Biomech. 40, 3178–3186. Richardson, J.B., Cunningham, J.L., Goodship, A.E., O'Connor, B.T., Kenwright, J., 1994. Measuring stiffness can define healing of tibial fractures. J. Bone Joint Surg. Br. 76, 389–394. Shefelbine, S.J., Simon, U., Claes, L., Gold, A., Gabet, Y., Bab, I., et al., 2005. Prediction of fracture callus mechanical properties using micro-CT images and voxel-based finite element analysis. Bone 36, 480–488. Sigurdsen, U.E., Reikeras, O., Utvag, S.E., 2009. External fixation compared to intramedullary nailing of tibial fractures in the rat. Acta Orthop. 80, 375–379. Smith, A., 2001. Laboratory Animal Science, first ed. Norwegian School of Veterinary Science, Oslo, Norway. Utvag, S.E., Grundnes, O., Rindal, D.B., Reikeras, O., 2003. Influence of extensive muscle injury on fracture healing in rat tibia. J. Orthop. Trauma 17, 430–435. Ward, K.A., Adams, J.E., Hangartner, T.N., 2005. Recommendations for thresholds for cortical bone geometry and density measurement by peripheral quantitative computed tomography. Calcif. Tissue Int. 77, 275–280.