SEARCH

SEARCH BY CITATION

Keywords:

  • 3D power Doppler;
  • high frequency ultrasound;
  • microCT microangiography;
  • microvasculature;
  • fracture healing

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We aimed to establish a novel approach with 3D high frequency power Doppler ultrasonography (3D-HF-PDU) to assess microvasculature at the fracture site in rat femurs by comparing with microCT-based microangiography. Twenty-four 9-month-old ovariectomized (OVX) osteoporotic rats and age-matched sham-ovariectomized (Sham) rats were used for establishing closed fracture models on right femora. At 2, 4, and 8 weeks post-operatively, four rats in each group underwent in vivo 3D-HF-PDU scanning for evaluation of vascularization and blood flow at the fracture site. Then the fractured femora were harvested for ex vivo microangiography, and neovasculatures within the callus were reconstructed for vascular volume analysis. Correlation between the vascular volumes of the two methodologies was examined. Both 3D-HF-PDU and microangiography showed a decline of vascular volume at the fracture site from 2 to 8 weeks and a significantly larger volume in the Sham group than the OVX group. A significant linear positive correlation (r = 0.87, p < 0.001) was detected between the volumes measured by the two methodologies. Osteoporotic rats had a diminished angiogenic response and lower blood perfusion than Shams. We believe 3D-HF-PDU is feasible and reproducible for in vivo assessment of microvasculature during femoral fracture healing in rats. © 2011 Orthopaedic Research Society Published by Wiley Periodicals, Inc. J Orthop Res 30:137–143, 2012

Angiogenesis plays a critical role in fracture healing especially in osteoporotic bone,1, 2 but the mechanisms that govern angiogenesis and their influences on impairing blood supply during the osteoporotic healing process remain poorly understood. The rat fracture model is widely used in fracture healing studies.3–5 However, quantitative techniques for studying vascularization, such as microCT-based microangiography, require animals to be euthanized and evaluated ex vivo.6–8 An in vivo, yet repeatable approach would allow quantitative measurements to serially monitor the revascularization at the fracture site and to study differences in microcirculation between fracture healing in normal and osteoporotic rats.

Power Doppler ultrasonography is based on the integrated Doppler power spectrum, which can be used to assess moving objects. In the vascular system, the power of the Doppler signals in relation to the number of moving erythrocytes can be detected and displayed in a color scale.9, 10 Due to its non-invasiveness and high sensitivity to slow flow, power Doppler in 2D was used to detect neovascularization in vivo with fracture healing.11–13 By using 3D power Doppler ultrasonography, reconstructed 3D vascular trees within a volume of interest (VOI) can be studied clinically. The vascularization can be objectively evaluated using specialized imaging software.14–17 However, due to the relatively narrow dynamic range of flow detection, the most frequently used clinical ultrasound frequency is unsuitable for studying slow flow in small animals. To improve detection capability and sensitivity, high frequency ultrasound is recommended for studying microvasculature in small animal models.18

To understand revascularization and local blood perfusion during normal and osteoporotic fracture healing, 3D high frequency power Doppler ultrasonography (3D-HF-PDU) was utilized and compared with microCT-based microangiography. Our aim was to validate the application of 3D-HF-PDU imaging for in vivo qualitative and quantitative assessment of microcirculation and angiogenesis in a small animal fracture healing model.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Animals and Closed Femoral Fracture Model

Twelve 6-month-old female Sprague–Dawley (SD) rats (220 ± 15 g) were bilaterally ovariectomized (OVX) using our established protocol19, 20 and housed for 3 months to induce osteoporosis. Another 12 age-matched rats underwent sham operations (Sham) following the same procedures with some surrounding visceral fat resected instead of the ovaries. All animals were obtained from the Laboratory Animal Services Center of the Chinese University of Hong Kong, and the procedures were approved by the Animal Experimentation Ethics Committee of the University (Ref: 08/004/ERG). A closed fracture was created at the right femoral shaft of each rat by using a three-point bending apparatus following intra-medullary insertion with a sterilized Kirschner wire (equation image1.2 mm, Sanatmetal Ltd, Eger, Hungary) following an established protocol.4, 5 Fracture was confirmed by anteroposterior and lateral radiographs. Rats in both groups were allowed free cage movements, and randomized to receive the following assessments at 2, 4, or 8 weeks post-fracture.

3D High Frequency Power Doppler Ultrasonography

At each time point, rats received 3D-HF-PDU assessment at the fracture site using Vevo-770 high frequency In Vivo Micro-Imaging System (VisualSonics, Toronto, Ontario, Canada). After general anesthesia (ketamine, xylazine, and 0.9% saline mixed in 3:2:3 ratio, 0.2 ml/100 g body weight), the rat was positioned prone on a flat, heated pad with body temperature maintained. The extremities were secured, and coupling gel was loaded to cover the exposed callus region.21 Two-dimentional real-time B-mode scanning was used to visualize the femur, and the scanning window was centered at the fracture line, with a 7.4 mm × 7.0 mm field of view. The ultrasound transducer (center frequency: 55 MHz), held by a hand-free stand, was positioned 4.5 mm above the central portion of the callus so as to match the focal zone. Then the device was switched to the 3D power Doppler mode (gain: 20 dB; pulse repetition frequency: 5 kHz; wall filter: 2.5 mm/s), and the scanning was constructed by a linear translation of the transducer along the femoral axis perpendicular to the single plane of 2D imaging. The translation rate was 0.05 mm/s. Based on experience, the movements of the leg from the heart beat, respiration, and the capillary pulsation were small, so electrocardiogram guiding for acquisition was not adopted. A scanning step of 0.05 mm was used with a scanning range of 10.0 mm. After lateral side scanning of the fracture site, the rat was turned to the supine position for medial side scanning site. Two hundred images were collected for each side within 30 min after induction of anesthesia. A rectangular region of interest (ROI) was manually outlined on the 100th slice by a single investigator. The ROI height was set from the exterior margin of the external callus to the medullary cavity, and the length was 7.40 mm (the length of scanning window). The same ROI selection was performed automatically in the remaining 2D slices by a custom-designed script of Matlab (Version7.0, The MathWorks, Inc., Natick, MA, Fig. 1). Vascular volume was evaluated by counting colored pixels on each slice at resolution of 0.016 mm × 0.016 mm, then multiplied to the 0.050 mm slice thickness. The vascular volumes from the medial and lateral sides were summed to obtain the total vascular volume. The customized script was used to discriminate and filter signals from large vessels; for an intra-linked region of flow signal displayed in the 2D image, the first criterion assessed is the area it covered. If the area was >1.3 mm2, a second criterion was set by observing the short axis of a regressed oval for this region. If the short axis was also larger than a critical value (0.35 mm), this region was judged as a large vessel and excluded in the calculation. Vascular volume at the fracture site was then evaluated as the sum of the colored voxel volumes of both sides. Mean signal intensity of blood flow was evaluated as the average amplitude of color voxels converted to a relative value from 0 to 255. The relative number of moving erythrocytes around the fracture site was then calculated by the formula: relative number of erythrocytes = (number of color voxels on the lateral side × mean signal intensity of at the lateral side) + (number of color voxels of the medial side × mean signal intensity at the medial side).

thumbnail image

Figure 1. Evaluations of the vascular volume and the mean signal intensity of blood flow from 3D-HF-PDU images. Rectangle in red involved the external callus. Upper and lower arrows show the mean vascularity percentage and the mean signal intensity within the VOI, respectively. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/jor]

Download figure to PowerPoint

Microfil Perfusion and Microangiography

microCT-based microangiography was performed following a previously established protocol.6–8 After the 3D-HF-PDU measurements, the abdominal aorta and inferior vena cava were exposed. Then the vascular system was perfused with Microfil MV-117 (Flow Tech, Carver, MA). The fractured femora were harvested, fixed in 4% paraformaldehyde, and decalcified by 10% EDTA (pH 7.4). The specimens were then subjected to VivaCT 40 (Scanco Medical, Brüttisellen, Switzerland) scanning according to our protocol.4 The scan range covered 3.7 mm proximal and distal to the fracture line. Three-dimensional vasculature images were reconstructed, and the vascular volume within callus was calculated.

Statistical Methods

The reproducibilities of the volume acquisition (n = 6) and the ROI selection (n = 10) were assessed by intra- and inter-observer intra-class correlation coefficients (ICC). Quantitative data of vascularization parameters were expressed as mean ± standard deviation. Differences in vascular volume and relative number of erythrocytes between normal and osteoporotic fracture at each time were compared using Mann–Whitney U-test. Correlation of vascular volume between 3D-HF-PDU and microangiography was analyzed by a two-tailed Pearson's correlation. All analyses were performed with SPSS software (version 16.0, SPSS, Inc., Chicago, IL). A p-value of <0.05 was regarded as significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

3D High Frequency Power Doppler Ultrasonography

Blood flow signals were present in the callus, periosteum, and peripheral soft tissues (Fig. 2). From the vascular volume assessment, a declining trend was found from 2 to 8 weeks. At each time, the Sham group showed larger vascular volumes than the corresponding OVX group. Quantitative analysis demonstrated a significantly larger value in the Sham than the OVX group at 4 weeks (p = 0.050, Table 1).

thumbnail image

Figure 2. Images of fractured femur in sagittal (a), coronal (b), and horizontal (c) planes. The color signals represent microvasculature around the callus. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/jor]

Download figure to PowerPoint

Table 1. Vascular Volume Evaluated by 3D-HF-PDU and Microangiography
Vascular Volume (mm3)Week 2p-ValueWeek 4p-ValueWeek 8p-Value
ShamOVXShamOVXShamOVX
  • *

    p < 0.05 between Sham and OVX.

3D-HF-PDU5.48 ± 0.214.74 ± 1.280.3864.27 ± 0.932.36 ± 0.610.050*1.45 ± 0.271.12 ± 0.380.275
Microangiography9.10 ± 1.136.59 ± 0.630.014*4.61 ± 1.212.89 ± 0.350.028*2.68 ± 0.811.90 ± 0.870.197

Signal intensity of blood flow in lateral and medial sides of femur increased during the early healing phase and peaked at 4 weeks; thereafter, signal intensity decreased. Between Sham and OVX groups at each time, osteoporotic rats showed a diminished, but insignificant intensity as compared to the shams (Table 2).

Table 2. Mean Signal Intensity of Blood Flow at the Fracture Site
Signal Intensity (0–255)Week 2p-ValueWeek 4p-ValueWeek 8p-Value
ShamOVXShamOVXShamOVX
Lateral side124.4 ± 4.5118.6 ± 10.50.480155.2 ± 37.5142.0 ± 36.50.827130.2 ± 7.3121.8 ± 10.20.275
Medial side123.1 ± 6.0116.8 ± 7.20.157156.8 ± 5.8141.6 ± 44.10.513128.8 ± 25.5109.1 ± 9.90.275

The relative number of erythrocytes at the fracture site decreased from 2 to 8 weeks. Comparing the two groups at each time, the erythrocyte number of osteoporotic rats at the fracture site was less than the shams. However, no significant difference was found (Table 3).

Table 3. Relative Number of Erythrocytes at the Fracture Site
 Week 2p-ValueWeek 4p-ValueWeek 8p-Value
ShamOVXShamOVXShamOVX
Relative number of erythrocytes (×107)4.67 ± 0.314.37 ± 0.290.0834.32 ± 0.752.45 ± 1.230.1271.47 ± 0.181.04 ± 0.310.127

All ICCs were >0.75 (Table 4), indicating good reproducibility.

Table 4. Intra- and Inter-Observer Reproducibility of Volume Acquisition and ROI Selection
Vascularization ParametersVolume AcquisitionROI Selection
Intra-ICCInter-ICCIntra-ICCInter-ICC
  1. Intra-ICC indicates intra-observer intra-class correlation coefficient and inter-ICC indicates inter-observer intra-class correlation coefficient.

Vascular volume0.9060.8590.9850.941
Mean intensity of blood signals0.8590.8260.9980.963

MicroCT-Based Microangiography

Vascular volume decreased gradually from 2 weeks onwards. Vascular volumes in the Sham group were larger than the OVX group at each time, significantly so at 2 and 4 weeks (p = 0.014 and 0.028, respectively, Table 1).

Correlation

The similarities of the 3D reconstructed images of power Doppler and microangiography (Fig. 3) confirmed that 3D-HF-PDU provided accurate assessments of microvasculature into the anatomical structure of the fracture site. By comparing vascular volumes obtained by the two techniques, we found a significant positive linear correlation (r = 0.87, p < 0.001, Fig. 4). However, no significant correlation was detected at individual times (2 weeks: r = 0.360, p = 0.381; 4 weeks: r = 0.463, p = 0.355; 8 weeks: r = 0.632, p = 0.179, Fig. 5).

thumbnail image

Figure 3. (a) Three-dimentional reconstruction of power Doppler images at the fracture site (blood vessels in red, callus in green and surrounding soft tissues in blue). (b) Microangiography within callus. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/jor]

Download figure to PowerPoint

thumbnail image

Figure 4. Correlation between the vascular volume by 3D-HF-PDU and by microangiography at the fracture site. A significant positive linear correlation was found.

Download figure to PowerPoint

thumbnail image

Figure 5. Correlations between the vascular volumes by 3D-HF-PDU and microangiography. A positive linear but insignificant correlation was found at each time.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Vascular volume calculated by 3D-HF-PDU showed a similar decline as that from microangiography from 2 to 8 weeks in Sham and OVX groups, similar to other studies of neovascularization in bone repair. For example, Raines et al.22 observed a decrease of neovascular volume in the marrow cavity from the peak value at 14 days after bone drilling in tibial marrow ablation rats. Zeng et al.23 found a high expression of vascular endothelial growth factor lasting from 14 to 28 days post-femoral fracture in rats; then it decreased gradually until 6 weeks.

We found an impaired angiogenic response in osteoporotic fracture healing. Other reports confirmed impaired revascularizaion in osteoporotic as compared to normal healing.1, 2 The mean signal intensity of blood flow is positively correlated to the concentration of moving blood cells.24 Risselada et al.'s data were similar to ours, in which the mean of signal scores of neovasculature, based on signal intensity peaked between 11 and 20 days post-fracture. A gradual decrease was seen thereafter.13 By utilizing the laser Doppler flowmetry, an optical in vivo imaging technique, Zheng et al.25 reported an increase in blood flow intensity from 2 to 4 weeks during bone defect healing in rabbits. The relative number of erythrocytes, calculated as the product of signal intensity and vascular volume, combines information of blood vessels and blood cells to evaluate local circulation.17, 26 We found a decrease of blood perfusion during callus mineralization and remodeling. The Sham group had better blood supply at the fracture site than the corresponding OVX group.

Due to the small sample size at each time, no significant correlation of vascular volume was found between 3D-HF-PDU and microangiography. However, all outcomes were pooled, a significant correlation was found. Our findings indicate 3D-HF-PDU is comparable to microangiography, and the results are consistent with previous studies. Also, 3D-HF-PDU is sensitive in distinguishing between normal and impaired angiogenic response. To our knowledge, this is the first study to assess changes in microvasculature and local blood flow during fracture healing in rats using 3D-HF-PDU. Currently, despite conventional microangiography being the gold standard for assessing microvasculature studies in fracture healing,6–8 it does not allow longitudinal analyses. Also, the results cannot provide flow information and can be affected by the quality of capillary perfusion due to fluctuations variables such as blood flush, perfusion pressure, and the threshold value chosen of the vascular tree during CT analysis.27 Conversely, 3D-HF-PDU is non-invasive, real-time, and can be used for longitudinal follow-up for both vascularization and blood flow quantifications.

In previous in vivo studies, investigators adopted power Doppler ultrasound to assess neovasculature at the fracture site in humans and large animals, detecting the existence of vascularity, evaluating the vessel area, vessel density, or blood flow intensity from 2D images.11–13, 28 However, since each 2D slice gives a different percentage of color pixels, many parallel slices must be measured throughout the total volume to reach a reliable result. Since 3D geometric evaluation can provide the spatial vascular tree, the value of vascular volume and blood flow within the VOI, 3D-HF-PDU measurement will improve the accuracy, objectivity, and integrity of the microcirculation information. Conventional frequency (2–15 MHz) ultrasound has difficulties in detecting small vessels (5–100 µm) and low flow velocities (∼0–50 mm/s). One feasible method is to raise the operating frequency to >20 MHz.29 Goertz et al.'s30 in vivo experiments confirmed that, at a center frequency of 50 MHz, the detection of vessels could be improved to 15–20 µm diameter in the mouse ear and demonstrated flow imaging in assessing a wide range of velocities (1–25 mm/s) in superficial mouse tumors. Therefore, we chose the transducer with the center frequency at 55 MHz, which was sensitive in detecting and assessing neovascularization around the fracture site.

There are some technical precautions for 3D-HF-PDU. First, the image noise must be reduced and filtered during scanning. Power Doppler is highly sensitive to motion artifacts that can be created by artery pulsations, breathing, or muscles contractions. An overly high Doppler gain, increasing the scan speed and vibration of the motor used in motor-steered probes may also generate disturbances. To reduce these artifacts, animals should be anesthetized with the extremities fixed, and a hand-free stand was utilized to mount the transducer for scanning. Also, it is appropriate to set the gain by up-regulating its value until random noise is encountered, then down-regulating until the noise disappears.31 The noise and low frequency flash artifacts could also be avoided by means of wall filters.32 Low filter settings can improve sensitivity, but may generate flash artifacts, whereas high filters can reduce the artifacts but will filter out flow in low levels. We found the wall filter at ∼2.5 mm/s to be suitable for scanning around the rat femur. Second, when dealing with the signals to assess microcirculation, appropriate imaging programs must be used to discriminate and filter large vessel signals. Finally, anesthesia will affect peripheral blood circulation, so each animal should be anesthetized with the same dose according to its body weight, placed at a set temperature, and positioned on a warming plate at that temperature to avoid variations caused by anesthesia. Complete scanning should be finished within a short time.

The application of 3D-HF-PDU has limitations. The detection of flow signals is limited by the penetration depth. With the use of higher Doppler frequency, penetration depth would be decreased, so researchers must compromise between accuracy and detection depth. We measured the muscle thickness at lateral and medial sides of the fracture site, which were 2.47 ± 0.33 and 4.06 ± 0.42 mm, respectively. The focal length of our transducer was 4.5 mm from the transducer surface in the soft tissues. Therefore, the resolution and penetration depth were adequate. Moreover, most sound waves were reflected as soon as they reached hard callus or cortical bone, so some vasculatures in the medullary cavity could not be detected, explaining why the absolute volume values detected by 3D-HF-PDU were consistently smaller than those of microangiography. There might be some overlap of the volume estimation depending on the size of the acoustic beam width. However, the beam width of a current ultrasound transducer is nearly identical to the scanning step size, so it will not significantly affect the comparison results.

Due to the advantages of small size (2–3 µm) and high backscatter that enhance the intensity of the signal, contrast agents have been developed to visualize vasculature at the capillary level, allowing a more sensitive measure of microcirculation.33, 34 We aimed to explore the feasibility of 3D-HF-PDU for assessing microvasculature, so we did not use contrast agents. However, for further improvement of image quality and more precise quantification, contrast agents are recommended for angiogenesis assessment at fractures in small animals.

In summary, our newly established technique using 3D-HF-PDU improved power Doppler 2D imaging to a 3D spatial view. Our results demonstrate the feasibility and reproducibility of 3D-HF-PDU for detecting and quantifying angiogenesis in fractured femurs of rats. Being non-invasive, while offering high-resolution anatomical visualization and objective data analysis, 3D-HF-PDU provides a robust approach for the evaluation of neovascular networks in the fracture healing of small animals, especially for in vivo longitudinal studies.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

This study was supported by Hong Kong General Research Fund (ref. no: 462708). The authors thank Dr. Fang-Yuan Wei and Mr. Wai-Ching Chin for their assistance in the power Doppler assessment and microCT analysis, respectively.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES