INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE MECHANISMS INVOLVED IN moderating skeletal adaptation are not clearly understood, but blood flow may play an important role. Currently available techniques to measure blood flow in bone, however, are time consuming and require destruction of the tissue. Laser-Doppler techniques have recently been introduced as less invasive methods to measure perfusion in soft and hard tissues. These techniques, laser-Doppler flowmetry (LDF) and laser-Doppler perfusion imaging (LDI), are based on the measurement of the frequency shift of incident photons after they collide with moving erythrocytes circulating through a vascular bed. LDF is an effective technique for measuring temporal variations in blood flow and monitoring dynamic responses to physiological stimuli at a specific site. LDI, in addition to monitoring changes in blood flow, however, offers the ability to study regional variations and heterogeneity in tissue blood flow and measure average tissue blood flow.

LDI has been validated for in vivo measurement of blood flow in soft tissues, where high correlation to microsphere-determined blood flow was found (3). LDI has yet to be validated as a tool for measuring bone blood flow, but its similarities to LDF and use in measuring perfusion in soft tissues suggest it may be a potential technique to measure changes in perfusion in hard tissues.

The purpose of this study was to determine the utility of LDI to measure perfusion in cortical bone by comparison of LDI output with colored microsphere (CM)-determined blood flow. The reproducibility of LDI output was also examined, and we determined whether near-infrared (NIR) and red wavelengths measured flow in the same region of bone. It was hypothesized that LDI was a feasible method to determine perfusion in cortical bone.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Twelve mature New Zealand White rabbits (6.0 ± 0.4 kg; Riemens Fur Ranche, St. Agathe, ON, Canada) were randomly assigned to one of three groups: normal control, constriction (norepinephrine), or dilatation (nitroprusside). Rabbits were premedicated with 0.15 ml intravenous acepromazine maleate and subsequently anesthetized with a mixture of halothane (2.5%) and 100% oxygen (1 ml/min). Body temperature was maintained at 37°C with the aid of a heating pad.

LDI. Blood flow at the cortical bone surface was measured with red (634-nm) and NIR (810-nm) wavelengths of the laser-Doppler perfusion imager (Moor Instruments, Devon, UK). The scanner head was placed ~25 cm above the exposed bone. Scanning speed was set at 10 ms/pixel, which gave a frequency bandwidth of 0.1-15 kHz. Flux and direct-current gains were adjusted to maximize signal strength while preventing saturation of the photodiode.

1

1

Evaluation of blood flow. CM blood flow determination was performed according to standardized protocol (7), with modifications for low-flow tissues (2). Briefly, a cannula was inserted into the left ventricle via the common carotid artery, and its placement was confirmed by a ventricular pressure waveform from an on-line pressure transducer. Arterial blood pressure was continuously monitored by the pressure transducer except during infusion of the CMs. A reference blood sample was withdrawn at a rate of 3 ml/min starting 5 s before infusion of the CMs through a cannula inserted into the left carotid artery. Approximately 10.2 million 15.5-µm CMs (Triton Technology, San Diego, CA) were infused into the left ventricle over a period of 30 s. The reference sample was continuously withdrawn for another 60 s after infusion of the microspheres, for a total collection period of 95 s. LDI measurements, at 810-nm wavelength, were obtained concurrently with CM determination. The animal was killed with an overdose of pentobarbital sodium. Final LDI scans were performed after death to obtain a "biological zero" reference point.

Blood flow determination. Kidneys and tibiae including the periosteum covering the scanned area were removed, weighed, and prepared for blood flow processing. The cortical bone of the medial middiaphysis was removed with a rotary cutting tool (Multipro Variable Speed, Dremel, Racine, WI). All traces of marrow were removed, and the samples were hand ground to ~1.5-mm thickness. Samples of cortical bone were ~51 mm × 6 mm and had a mass of 1.2 ± 0.2 g. Reference blood samples and kidneys were digested in 10 ml of 4 M KOH at 60°C for 24 h. Bone samples were decalcified in 10 ml HNO₃ for 3 days at room temperature and then digested in 10 ml 4 M KOH at 60°C for 2 days. The digested tissues were filtered through 8-µm filters (25 mm, Nucleopore Track Etch Membranes, Whatman) and either directly counted using an epifluorescent microscope (periosteum and bone) or counted by spectrophometry (reference samples and kidneys). Standardized blood flow (ml · min¹ · 100 g¹) values were determined by relating tissue to reference blood CM counts and then normalizing to the sample mass (2).

Statistical methods. Normality of the data was determined by a one-sample Kolmogorov-Smirnov test. Differences between right and left perfusion values, and between pre- and postdrug infusion perfusion values, were compared using Student's paired t-tests. Coefficients of variation were determined to estimate differences between repeated measures of LDI output signals. Correlation between red and NIR wavelengths was determined by linear regression. Correlation between LDI perfusion values and bone blood flow measured by CM technique was analyzed using a nonparametric two-tailed Spearman's rank correlation. A level of P < 0.05 was used to detect significant differences.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sample images generated by the LDI using red and NIR wavelengths before drug injection, immediately after a 3-min infusion of the drug (NIR), and postmortem (NIR) are depicted in Fig. 1. LDI revealed heterogeneity in the tissue. One rabbit was removed from the study after examination of the tibial cross section because of inconsistencies in the bone. One sample of one rabbit was also lost during the filtering stage of the microsphere testing and thus deleted from the analysis.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Sample laser-Doppler perfusion imaging (LDI) images (from left to right) before drug injection [normal red and normal near-infrared (NIR) wavelengths], after drug infusion (NIR), and postmortem (NIR). The scale represents the spectrum of perfusion units (PU) for each wavelength.

Injection of nitroprusside and norepinephrine resulted in approximately a 40-50% decrease or 20-30% increase, respectively, in blood pressure. Both drugs significantly changed perfusion values compared with baseline.

Mean perfusion values of each limb at red and NIR wavelengths were normally distributed. No significant differences existed between mean perfusion values of the right and left limbs measured with red or NIR wavelengths. Consecutive LDI perfusion measurements of each bone resulted in an average coefficient of variation of 0.05 (range 0.01-0.14).

The normal (before drug application) perfusion values for normal control, dilatation (nitroprusside), and constriction (norepinephrine) groups at red wavelength were 16.4 ± 5.6 (SD), 22.7 ± 9.8, and 17.1 ± 6.7 PU, respectively. At NIR wavelength, the perfusion values were 28.3 ± 10.1, 36.6 ± 11.3, and 30.3 ± 11.3 PU. Figure 2 compares the mean LDI output signal from the red and NIR lasers. A slope of 1.31 NIR PU/red PU and y-intercept of 7.28 PU were determined (r = 0.93, P < 0.05). Figure 3 compares the average standard deviation in the images, as measured by the LDI, from the red and NIR lasers. A slope and y-intercept of 1.18 NIR PU/red PU and 5.71 PU, respectively, were determined (r = 0.94, P < 0.05).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Relationship of NIR vs. red wavelength mean output. LDI output of the 2 wavelengths were significantly correlated (r = 0.93; P < 0.05).

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Relationship of NIR vs. red wavelength standard deviation (std) output. LDI output of the 2 wavelengths were significantly correlated (r = 0.94; P < 0.05).

Average blood flow measured by CMs for normal control, dilatation (nitroprusside), and constriction (norepinephrine) groups were 0.68 ± 0.26, 0.51 ± 0.36, and 0.39 ± 0.36 ml · min¹ · 100 g¹, respectively. At NIR wavelength the LDI perfusion values were 24.5 ± 13.0 PU (normal control), 14.1 ± 3.7 PU [dilatation (nitroprusside)], and 16.9 ± 4.4 PU [constriction (norepinephrine)]. There were no significant differences in CM-determined blood flow between normal control, dilatation (nitroprusside), and constriction (norepinephrine) groups. The average number of CMs per bone sample in normal control, dilatation (nitroprusside), and constriction (norepinephrine) groups were 70.8 ± 23.2, 41.2 ± 23.3, and 35.9 ± 26.7, respectively.

A significant correlation (r = 0.6, P < 0.05) was detected between LDI perfusion values and CM-determined blood flow (Fig. 4). All data were linearly related (P < 0.05).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Relationship between mean LDI output (perfusion units) and standardized blood flow (measured with colored microspheres). A significant correlation between LDI and colored microsphere-determined flow was found (r = 0.6; P < 0.05)

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

LDI has been shown to be able to detect changes in perfusion in bone (4), and its output has been significantly correlated to CM-determined flow in ligaments (3). Here, we ascertained the utility of LDI to measure cortical bone perfusion by examining its reproducibility, dual-wavelength capabilities, and comparison with a known standard measure of blood flow. LDI measurement of perfusion in cortical bone was highly reproducible and significantly correlated with CM-determined blood flow. LDI showed sufficient sensitivity to detect relative physiological magnitude changes in blood flow created by infusion of vasoactive agents. Red and NIR wavelengths of the LDI appeared to be sampling flow in the same volume of bone.

It is advantageous to use NIR wavelengths because they are better able to penetrate through highly pigmented tissues. In stratified tissues such as skin, however, use of different wavelengths may allow the measurement of different regions of flow (1). The high correlation and significance of the linear relationships between red and NIR wavelengths suggested that the two wavelengths were measuring the same regions of blood flow in cortical bone or that blood flow was similar in the region measured by NIR compared with that measured by red wavelength (5). The high correlation between the standard deviations of the images suggested that the two wavelengths were measuring a similar flow region and/or the heterogeneous structure in the upper layer of bone measured by red wavelength was highly correlated to the lower layer measured by NIR (5).

Even after subtraction of the biological zero from the mean perfusion values, the nonzero y-intercept from the comparison of CM-determined flow and LDI output was detected as in previous studies (5). The intercept was likely a combined effect of the measurement process and the instrument's inherent offset (5). Laser-Doppler systems often have an inherent signal offset that would result in a positive value for a no-flow condition (3). Because of the photon direction insensitivity and diffusive light scatterings in laser-Doppler technologies, these instruments give an output signal higher than electronic zero when the net blood flow in tissue is apparently zero (18). The Brownian motion of particles, random wandering movement in a no-flow situation, although not related to organized blood flow, may also be detected by LDI and affect perfusion values. Subtraction of the biological zero should have eliminated these two effects, suggesting averaging LDI pixel values may have introduced a statistical artifact and resulted in a nonzero intercept. The signal processor of the LDI gave a linear output for a limited range of flows, but the frequencies outside of this range were filtered out (3). The curve for low flows may be nonlinear, and thus the true relation may contain higher order terms that would force the curve to pass through the origin (5). The comparison between LDI and CM may have also accounted for the nonzero y-intercept, because neither method is a "true" measure of flow nor accurate for measuring extremely low flows. The overlap in standard deviations of LDI and CM data may also have been due to the less accurate measurement of low flows and the intra-animal variability.

The reproducibility of multiple measurements of the same area, including realignment of the LDI after scanning the opposite leg, was within 5% of the other scans. Although the coefficient of variation varied between 0.01 and 0.14 (mean = 0.05), the reproducibility of LDI scans was higher than multiple LDF measurements in the same position in cancellous bone (mean coefficient of variation between 0.05 and 0.15) (9, 15). Because of the spatial variation in cancellous bone, however, the coefficient of variation at four different positions was between 0.09 and 0.21, and, although not significantly different, single-point measurements were not representative of an entire tissue (8).

LDF is a recognized technique for measuring temporal variations in blood flow and monitoring dynamic responses to physiological stimuli at a specific site, but LDI offers the ability to study regional variations in tissue blood flow, image tissue heterogeneity, and measure average tissue blood flow. Because of the heterogeneity of small-vessel distribution in vascularized tissue, measurements made by laser-Doppler flowmeters are site specific (14). Although both LDI and LDF require exposing of the tissue of interest, the necessity of contact between the LDF probe and tissue may introduce the risk of infection and induce a motion artifact in the LDF signal. Although a linear relationship was attained between LDF output and controlled perfusion rates to an isolated tissue, the slope of the relation varied among animals (13) suggesting the results could not be compared for different animals. Smits and colleagues (14) showed regional variation in tissue blood flow, and that by averaging six to eight points across the tissue surface, LDF was linearly related to radioactive microsphere-determined blood flow. This method, although a better assessment of tissue blood flow than a single point, was subject to temporal variations in tissue blood flow over the period of time required to move the probe to obtain sufficient data points to represent accurately the entire tissue (3).

Laser-Doppler technologies are advantageous compared with most methods of blood flow determination because they allow for continuous measurement of flow without tissue contact. Although Swointkowski and colleagues (15) did not find significant correlation between the estimation of bone blood flow in rabbits by the microsphere method and LDF (15), Lausten and co-workers (9) were able to obtain significant correlation between LDF output and flow measured by microspheres. The discrepancy in experimental results comparing LDF and microsphere flows may have been due to the area in which the average bone blood flow was measured by the microsphere technique was considerably larger than the area measured by LDF (9). Similar to LDF where LDI measures flow over a period of time, comparison to microsphere-derived flow at one specific time point may have accounted for some of the differences between the results of bone blood flow measurements from the two techniques in the present study.

Although it has been suggested that to obtain an accurate estimate of bone blood flow each sample should contain 150-250 radioactive microspheres (10), the CM-determined flows in this study were similar to previous studies of the tibia in rabbits (16, 17). There was a 14% error if the bone sample contained <50 microspheres; however, the relative error dropped to <10% in samples containing >150 microspheres (10). Because LDI has the capability of measuring flow in any given area, the area in which average bone blood flow was measured by the microsphere technique was the same as that measured by LDI.

Laser-Doppler techniques can only provide a relative estimate of microvascular perfusion, and, although it may be difficult to calibrate the signal into absolute values because of the variation in optical properties of various tissues and spatial variation in the microvascular bed (9), previous studies suggest it may be possible to derive calibration factors for a specific tissue from LDI output (3). A calibration factor to convert laser-Doppler output into absolute units of flow has been previously investigated using LDF (14). A single calibration factor could not be derived for all tissues because the optical properties of different tissues required a different calibration factor for laser-Doppler measurements (3). The LDI output signal cannot be calibrated into absolute values, but its reproducibility and the averaging to compensate for tissue heterogeneity suggest that coefficients for a specific tissue could be derived (3).

The present study suggested that laser-Doppler perfusion imaging is a promising tool for imaging in vivo changes in perfusion in cortical bone. Although use of LDI may be limited to measurements on and near the bone's surface, similar to LDF (11), it could be beneficial for studying the influence of blood flow on bone remodeling where areal perfusion is of importance.