| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Cardiovascular Physiology, 2Institute for Anatomy II, Heinrich-Heine-University, 40225 Düsseldorf, Germany; and 3Laboratory of Cardiac Energetics, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland 20892
Submitted 5 February 2004 ; accepted in final form 23 April 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
myocardial perfusion; microscopy; fluorescence; magnetic resonance; iron oxide particle
For microsphere deposition density to be a valid measure of local arterial organ perfusion, the central assumption is that a homogenous distribution of microspheres in the aortic outflow results in a delivery of microspheres to the organ or tissue region that is proportional to and solely dependent on flow. Provided that a complete extraction in the first pass through the organ and a sufficient temporal and spatial stability of the marker are given, the deposition density in a tissue or organ is governed by flow and probability only (see also Ref. 22). Clearly, today's commercially available microspheres are temporally stable and will reside in the tissue for days to weeks (28). These microspheres are nearly completely extracted during the first pass and appear to be well mixed in the aortic outflow after being adequately sonicated and vortexed. However, the behavior of these rigid microspheres, twice the size of erythrocytes, in the microcirculation has always been a reason for concern. Whereas microsphere extraction in a given arteriolar or capillary vessel may result in blockage of the vessel (8), both enhanced collateral circulation and incomplete obstruction (described as a "poorly fitting ball valve with individual blood cells squeezing past") (12) will reduce its impact. In addition, in two studies, Bassingthwaighte et al. (3, 4) demonstrated reassuringly a close correlation between the deposition density of radioactive microspheres and a nearly ideal molecular marker (2-iododesmethylimipramine) in sheep and rabbit hearts. In these studies, microsphere deposition density exceeded the deposition density of the molecular marker only at high local flow rates, suggesting some streaming artifacts only in these high-flow regions. A close correlation between microspheres and another molecular flow marker, adenosine, was recently also reported in canine hearts (16). Therefore, microsphere deposition density in principle is a valid measure of local myocardial perfusion.
Because it is based on the measurement of a finite number of microspheres, the precision of the measurement is dependent on the number of microspheres per sample (6). At >400 microspheres in a given sample, the flow determined will be with 95% confidence within 10% of the true flow. While the precise measurement of flow in an individual sample therefore requires a relatively high number of microspheres, fewer spheres will be sufficient to estimate the variability of flow within a given organ when several samples are studied (21).
When local myocardial perfusion in individual samples of 0.25 to 1% of the left ventricular mass is studied, a substantial spatial variability of local myocardial blood flow is observed (spatial heterogeneity of perfusion) in several species with the use of either microspheres or molecular markers for flow determinations (e.g., Refs. 3, 4, and 15). In addition, a considerable body of evidence indicates a close correlation of local perfusion and metabolism as well as gene and protein expression (for a review, see Ref. 9). In these studies, the sample size investigated rarely dropped below 0.25 ml, because at a higher spatial scale, the same precision of the flow determination would require a high number of microspheres, possibly interfering with flow regulation and oxygen supply. For the same reason, the spatial variability of local myocardial perfusion in small laboratory animals has not been studied by microspheres, but only by imaging the retention of a molecular flow marker by autoradiography (26).
While recent advances in imaging techniques enable the accurate detection and localization of individual microspheres in the heart at high spatial resolution (5), the validity of the microsphere approach at a high spatial resolution (e.g., 1–50 µl) has not been rigorously tested. Specifically, it is unclear whether the central assumption of the microsphere technique (the deposition density is governed solely by local flow and probability) is valid at all spatial scales. We therefore studied the pattern of microsphere deposition at the highest possible resolution using epifluorescence microscopy, fluorescence imaging coupled to a cryomicrotome, and MRI of MRI-visible microspheres. The results clearly indicate that at a microscopic scale, microsphere deposition is not governed solely by flow and by chance. Therefore, at a high spatial resolution (e.g., <2 µl), microspheres will not be a valid measure of local flow.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Experimental procedures.
In the open-chest studies, anesthesia was achieved by volatile anesthetics (isoflurane, halothane, or enflurane, n = 12) or 
-chloralose (n = 5) (100 mg/kg iv loading and 50 mg/kg iv maintenance as required) combined with opioid analgesics (fentanyl). In ventilated beagle dogs (10–12 kg), following the onset of anesthesia, a midline thoracotomy was performed, the pericardium was removed, and the left atrium was cannulated for later application of microspheres. A femoral artery catheter enabled collection of blood from the abdominal aorta as an arterial reference organ. In a subset of experiments, a right branch of the left anterior descending coronary artery was cannulated for regional infusion of adenosine (0.7 mmol/l at 1 ml/min, approximate final concentration 5–10 µmol/l) for 20 min. At the end of the experiment, the animals were euthanized with KCl (10–40 meq rapid iv bolus) and the heart was excised. In one set of experiments, three different-colored fluorescent microspheres (15 µm, 250,000/kg) were given every 20 min, whereas in a second set of experiments, microspheres were applied before, twice during adenosine, and once thereafter.
In a further set of experiments, the deposition pattern of iron oxide microspheres (15 µm) was imaged by magnetic resonance imaging in the open-chest animal in vivo and after cardiac excision in vitro (see below). High-resolution in vivo imaging of iron oxide microspheres required the careful positioning of a surface coil on the myocardial surface of the left ventricular free wall, and the placement of the ventilated, open-chest animal in the MRI scanner. Control images were obtained before the animal was withdrawn from the scanner for application of the iron microspheres (106 microspheres/kg). Afterward, the animal was put back in the identical position it had taken up before.
To test the temporal stability of the myocardial microsphere distribution pattern in the awake dog, and to determine capillary volume and arteriolar density in areas of different basal flow, dogs weighing 20 kg were instrumented after an overnight fast, similar to previously published procedures (17, 20). Left atrial and aortic catheters were placed under general anesthesia for later measurements of flow by microspheres (11). The animals were allowed to recover for 9 days. In one dog, differently colored fluorescent microspheres (15 µm, 250,000/kg) were then applied on days 1–3 via the atrial catheter while blood was simultaneously withdrawn from the aorta (10 ml/min) into a virtual reference organ. For cardiac excision, the animal was anesthetized as previously described (10). In the other dog, radioactive microspheres were applied for local flow determinations in the alert animal. Cardiac excision was followed by glutaraldehyde fixation for electron and light microscopy analysis, as described previously (25).
In all of the other experiments, after cardiac excision the coronary arteries were flushed with a Ca2+-free medium composed of (in mmol/l) 280 sucrose, 10 HEPES, 1 EDTA, and 1 EGTA, pH 7.2. Atrial tissue was trimmed away and the heart gently placed in agarose (1%, 42°C), which solidified after cooling.
Fluorescent microspheres: application and imaging.
In a first set of experiments in the open-chest dog, yellow-green (Molecular Probes), orange-high, and violet-high (Stason IMT) and crimson (Molecular Probes) microspheres were applied in 20-min intervals. After cardiac excision, transversal (short axis) myocardial slices (
4 mm thickness) were obtained, and microsphere deposition was imaged with the use of a fluorescence stereomicroscope (model MZ FLIII, Leica Microsystems) and a green fluorescent protein filter.
In a second set of experiments, yellow, red, and scarlet microspheres (Molecular Probes) were applied in 40-min intervals, and microsphere deposition was imaged using a commercially available imaging cryomicrotome (Barlow Scientific; Olympia, WA) (5, 14). In brief, the microtome sectioned through the frozen heart at a selected slice thickness of 30–40 µm. Fluorescent images of the newly exposed myocardial surface were digitally obtained with appropriate excitation and emission filters to isolate each of up to four different fluorescent colors. At a 2,000 x 2,000 pixel resolution, the chosen field of view resulted in an individual pixel size of 30 x 30 µm2. As described in detail before (5), software analysis of the acquired images resulted in a list of all microsphere-containing voxels with their x- and y-coordinates according to the respective pixel size and z-coordinates according to the respective slice thickness. Because the pixel size exceeded the size of an individual microsphere, the algorithm employed would have been unable to identify clustering or clumping of microspheres within a given voxel (which might include in a worst-case scenario up to 8 microspheres, based on the relative sizes of the spheres and the voxels).
In the experiment studying microsphere distribution in the awake animal, yellow microspheres (Molecular Probes) were applied on day 1, red and scarlet microspheres simultaneously on day 2, and green microspheres on day 3. Finally, in the experiments focused on regional elevation of flow by adenosine, red and scarlet microspheres were applied sequentially during adenosine, whereas yellow and green microspheres were applied before and after adenosine. Imaging was carried out as described above.
Analysis of microsphere deposition pattern. To describe in a quantitative manner the pattern of microsphere deposition, a program was developed in IDL (Interactive Data Language, version 5.5, Research Systems; Boulder, CO). This program calculated for each individual microsphere containing voxel (identified by x-, y-, and z-coordinates from the image analysis described above) in a defined range of slices or volumes the distance to its nearest neighbor of a different color (intercolor distance) or of the identical color (intracolor distance). The frequencies of all individual next neighbor distances were thereafter displayed as a histogram with the minimal voxel dimension (slice thickness) as bin size.
A random distribution of microspheres in space at a defined density was simulated with the use of the random number generator routine of IDL 5.5. The density of these virtual microspheres was taken from the individual experimental observations. It was thus possible to compare the experimentally observed distance distribution (see above) with a randomly generated microsphere localization pattern of an identical microsphere density.
Iron oxide microsphere MRI. In a clinical 1.5-T MR scanner (Sonata, Siemens; Erlangen, Germany), cardiac pacing (120 beats/min) and ventilation (60 breaths/min) were synchronized to avoid movement artifacts. After localizers and a two-dimensional (2-D) flash cine imaging sequence, end-systolic T2-weighted control images were obtained by a 2-D gradient echo GRE sequence with dark blood preparation. The imaging parameters for the sequence were 128 x 128 mm2 field of view, 128 x 128 acquisition matrix, 2 mm slice thickness, and 1 k-space line/heart beat. The images were obtained at an echo time of 5, 10, 15, 20, and 25 ms in a single slice. As described above, the animal was withdrawn from the scanner before application of iron oxide microspheres. It was returned to its original position, before a second set of images was obtained.
After cardiac excision, higher-resolution images of iron oxide microsphere deposition patterns were obtained from the heart placed in agar. With the use of a three-dimensional (3-D) version of essentially the same sequence as before, the acquisition matrix was increased to 256 x 256 and the slice thickness reduced to 0.5 mm, resulting in an isotropic voxel size of 0.5 x 0.5 x 0.5 mm3. Repetition time was increased to 1 s. To cover the whole ventricular myocardium in a true 3-D mode, 96 contiguous slices were obtained per heart.
Chemicals.
Fluorescent microspheres (15 µm) were obtained from Molecular Probes (Eugene, OR) and IMT Stason (Irvine, CA). Iron oxide microspheres (15 µm, 0.5 ng iron oxide/microsphere) were also obtained from IMT Stason. All other chemicals employed were of analytic grade.
Statistics. The arrays describing the intra- and intercolor microsphere distances (see Analysis of microsphere deposition pattern) were statistically analyzed with the use of statistical software (version 11.0.1, SPSS, Chicago, IL; www.spss.com), calculating both mean and median. Also, the mode of each distribution of minimal distances binned by the smallest voxel dimension was identified as well as the second mode in case of a bimodal distribution.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
3,000 mm–2) and low number of microspheres per tissue volume (2.5 mm–3) (for details, see DISCUSSION).
|
31 µm thick were removed sequentially from base to apex. Fluorescent images were obtained of the sequentially exposed surfaces for each fluorescent color, resulting in a set of 1,200 images per heart and color. Representative images from one heart are presented in Fig. 2. Each panel demonstrates the fluorescence of one microsphere color only and displays several characteristic features. First, the deposition pattern on a qualitative level clearly appears to be nonrandom. Second, clustering of microspheres (highlighted by circles) is readily apparent, which may even take on the form of streak-like patterns. Third, contiguous spaces devoid of microspheres are present that may extend to several millimeters (highlighted by arrows).
|
3,000 mm–2), consistent with a previous report from our group (25). In contrast, the arteriolar density in the same samples as estimated by light microscopy (57 ± 34 mm–2) displayed a much greater spatial heterogeneity (CV 0.59). Because the high capillary density, the arteriolar branching pattern (13), and the low number of spheres applied were known, the similarity of the deposition pattern of sequentially applied microspheres was surprising (Fig. 2). In Fig. 2, each panel represents the deposition of one fluorescent batch of microspheres, where the yellow microspheres had been applied first, followed by the red and scarlet microspheres 40 and 80 min later, respectively. It is readily apparent that areas devoid of microspheres after the first application remained frequently devoid of spheres also after the second and third application (see arrows). Also, areas receiving a number of spheres, a cluster, on the first application, frequently also received several spheres on the second or third application, with the local pattern remaining constant (see circles). This stable deposition pattern on subsequent microsphere applications corresponded well with the observation of close proximity of sequentially applied microspheres as seen in Fig. 1. Please note that the microspheres imaged were approximately within 200 µm depth from the myocardial slice surface.
Quantitative analysis of fluorescence microsphere deposition pattern. To verify the qualitative impression of close proximity of sequentially applied microspheres and of a nonrandom deposition pattern, the position (in x-, y-, and z-coordinates) of each MS containing voxel in the heart was determined at a 31 x 30 x 30 µm3 resolution. This was made possible by the many high-resolution images at a small slice thickness, which enabled the application of a software algorithm that identified the voxel position (see METHODS). It was thus possible to calculate the distance of a given reference microsphere to its nearest neighbor of a different or the same color, enabling a thorough quantitative description of the noted microsphere clustering. This was done for all microspheres in all hearts, and the results expressed in frequency histograms as shown in Fig. 3.
|
25% of spheres may have been associated with the first mode, whereas most microspheres appeared to be within the second mode of the distribution. A bimodal distribution was observed in all hearts (see Table 1), and for each pair of microspheres, the first mode of the distribution was much smaller than the median of the total, i.e., between 10% and 33% of the median. Similarly, in all hearts the fraction of microspheres displaying a distance to its nearest neighbor smaller than or equal to the first mode was relatively large (3.3–19.9%), compared with the distance between the first mode and mean or median. In contrast, the second mode was in most hearts similar to or slightly larger than the median.
|
Comparison of the nearest neighbor distances of the empirically observed intercolor microspheres and of two sets of in silico, virtual microspheres also revealed a second aspect of the unexpected microsphere behavior. Whereas with randomly spaced microspheres only very few MS would have been seen at a distance greater than twice the median or mean (1 mm in Fig. 3), >8% of the empirical microspheres were seen at a distance >1 mm apart from its nearest neighbor of a different color. While the bimodal distribution is most likely due to the clustering and high temporal stability of the microsphere deposition pattern, the relatively high fraction of spheres at a greater distance most likely reflects the large voids of microspheres in the images.
The unexpected deposition pattern was not only seen when studying the distances between sequentially applied microspheres but also within a given microsphere color (batch). For microspheres randomly distributed in space, the distance between an index microsphere and its nearest neighbor could be expected to be close to a Gaussian distribution. In fact, when calculating for eight randomly distributed sets of virtual in silico microspheres the nearest neighbor distances, mean, median, and the mode of the distributions were similar (516 ± 36, 509 ± 35, and 468 ± 58 µm), the skewness was very small (0.26 ± 0.018) and the kurtosis negligible (–0.046 ± 0.039), all very much consistent with a Gaussian normal distribution. In contrast, for eight empirical sets of microsphere containing voxels obtained from four hearts, the distribution was always positively skewed to the left, indicated by a skewness ranging from 0.65 ± 0.01 to 1.185 ± 0.01 (0.95 ± 0.21; P < 0.001 vs. random distribution) and the mode being smaller than median and mean (380 ± 79 vs. 436 ± 24 and 473 ± 20 µm). Because the in silico microspheres were matched to the deposition density of the observed spheres, it is also noteworthy that the randomly distributed microspheres displayed a significantly greater median distance (509 ± 35 vs. 436 ± 24 µm; P < 0.001) than the empirically observed, again reflecting the microsphere clustering.
To test whether the high temporal stability of the microsphere deposition pattern extended from tens of minutes to days, and to study, whether the anesthesia present in open-chest animals might interfere with the local flow and microsphere localization, one animal was chronically instrumented and microspheres were applied starting 9 days after catheter placement. Aortic end-diastolic and peak systolic pressure were 88 and 148 mmHg, respectively, and remained stable during the following 2 days (93 and 161 mmHg on day 3); heart rate varied from 89 to 124 beats/min. As is readily apparent from Fig. 4, again a high fraction of spheres was much closer to its nearest neighbor than expected from two randomly spaced sets of in silico microspheres. It is remarkable that the distance distribution of two batches of microspheres given on day 3 and 1 was almost identical to two batches given simultaneously on day 2 (Fig. 4). Thus the deposition pattern previously documented in the anesthetized open-chest dog is also present in the chronically instrumented, awake animal.
|
|
MR imaging of microsphere deposition.
Because the deposition pattern of fluorescent microspheres can only be analyzed post mortem, we explored the possibility of MR visible microspheres for in vivo visualization of microsphere localization. Commercially available 15 µm iron oxide microspheres, containing 0.5 ng iron oxide each, were applied to the left atrium. Preliminary agarose phantom experiments had indicated that an iron oxide concentration of 2 µg/ml was sufficient to substantially decrease the image intensity in a T2-weighted gradient echo imaging sequence. In consequence, a medium number of microspheres (e.g., 1,000 MS/g, i.e., 0.5 µg FeO/g) was expected to be too low to significantly alter the imaging contrast in vivo and in vitro. This was confirmed by preliminary canine experiments. Therefore, the amount of microsphere applied was increased by a factor of four compared with fluorescent microspheres. This enabled the visualization of microsphere deposition in a myocardial slice at a voxel size of 2 µl, as demonstrated in Fig. 6. Whereas the top left panel depicts the slice before microsphere application in a T2-weighted flash image, after microsphere application a heterogeneous pattern of microsphere deposition was revealed. Areas apparently devoid of microspheres and streak-like features characterized this pattern, resembling the pattern seen with fluorescent microspheres before. After cardiac excision, in vitro imaging enabled the acquisition at a resolution of 125 nl (0.5 x 0.5 x 0.5 mm3). Two representative slices from 3-D data sets are displayed in the bottom two panels. Again, apparently microsphere-free areas and similar clustering features could be seen. In addition, a higher subendocardial microsphere density was observed.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Nonrandom localization of microsphere deposition.
The observed nonrandom localization is surprising, because an approximately random behavior of microspheres was an implicit assumption in all applications of microspheres for local flow measurements. In fact, microsphere deposition had even been formally treated as a mathematical problem that allowed statements regarding precision and distributions predominantly based on simulations (21). Also, because microspheres are homogenously distributed within the aortic blood stream and arrive at the organ of interest in proportion to its relative share in cardiac output, this approach appears to be justified. In this generally accepted model (8, 22), application of, e.g., 2.5 x 106 microspheres in a 10-kg dog may result in 125,000 microspheres, i.e., 5% of the total, arriving at the heart. At an organ mass of 50 g, 2,500 microspheres will be transported to a tissue volume element of 1 g. The total number of capillaries in this volume element may be 3.8 x 106, branching from 1.2 x 106 precapillary order 1 (9 µm) or 0.54 x 106 precapillary order 2 (13 µm) vessels (1, 13). There will thus be 1 microsphere per 200 second order, 500 first-order vessels or 1,500 capillaries. Assuming a homogeneous perfusion of a given tissue plane, the final destination of a given microsphere, i.e., the vessel extracting it, in this ideal scenario would only be governed by chance. After sequential application of three sets of microspheres, the likelihood of three individual microspheres being extracted in the same vessel would be very small [<(1/200)3, i.e., 1 in 8 x 106]. However, three sequentially applied microspheres were frequently observed within one apparent vessel (Fig. 1, a–c), clearly suggesting that the distribution of microspheres among all vessels within an imaging field was not only governed by the laws of probability. This observation was even more surprising because conventional expectations (8) predicted, in fact, contrary to anecdotal evidence (12), that a vessel following extraction of a microsphere might be blocked to further perfusion, and therefore not be able to receive a second microsphere thereafter.
The nonrandom character of microsphere deposition was further demonstrated by the spatial pattern of microsphere localization in myocardial slices, which was temporally stable for days (Figs. 2– 4). It was also underlined by the frequency of sequentially applied microspheres being extracted by the same myocardial volume element as defined by the spatial resolution of the imaging cryomicrotome system and the subsequent analysis (Fig. 3).
What could be the reason for this nonrandom distribution? Already when applying one set of microspheres, the deposition pattern seen, e.g., in Figs. 2 and 6 was nonrandom, characterized by clusters, but also by large, microsphere-free voids (Figs. 2 and 5). This was also indicated by a distance distribution skewed to the left and by a mode being considerably smaller than the mean (see RESULTS). Several factors may contribute to this effect: A nonhomogeneous distribution of microsphere lodging sites, i.e., small (15 µm) arterioles, which is indicated by our data on arteriolar density and consistent with recently published data (see, e.g., Ref. 24), may be one decisive factor. Also, arterio-venous fistulae in the microcirculation may contribute to an inhomogeneous distribution. Finally, spatial heterogeneity of perfusion will also contribute to this effect, although high-resolution measurements of microsphere deposition in canine hearts indicated a rather homogeneous pattern at this microliter scale (19).
However, given the small number of microspheres and the large number of potential lodging sites (see above), these factors do not explain why, after repeated applications, the microspheres finally lodge in the very same vessels. Because the transport characteristics of rigid 15 µm spheres in the microcirculation will deviate more and more from that of plasma or erythrocytes, we suggest that at branching points of medium or small coronary arterioles (e.g., <50 µm) rigid microspheres display considerable streaming phenomena. This is supported by the fact that multiple injections of FMS resulted in stacking of spheres in the same vessel implying that although the MS was "trapped", the flow, or delivery of spheres, was not reduced to zero in the vessel. These streaming artifacts were previously recognized in rather large areas of high flow, ranging from 1 to 0.25% of left ventricular mass (3, 4), and are consistent with in vitro experiments demonstrating a bias of spherical particles at branch points toward the higher flow branch (7). Consequently, for small individual sample sizes, supplied by small arteriolar vessels, microspheres will not represent a valid means for flow determinations.
Spatial distribution and scale of perfusion and microsphere deposition pattern. In contrast, when perfusion of organs and large tissue regions is studied, all available evidence indicates the validity of the microsphere approach (22). The close correlation between plasma flow measured by molecular flow markers and the blood flow determined by microspheres in individual samples as small as 50 mg in rabbits and 250 mg in sheep and dogs (3, 4, 16), clearly supports the validity of local perfusion measurements by microspheres for samples as small as 0.25–1% of the left ventricular (LV) myocardial mass. Dividing the LV myocardium into 4, 8, 16, ... , 256 individual pieces revealed spatial variations of regional flows within the heart (spatial heterogeneity of flow) that could be described by a fractal pattern in the spatial range observed, with increasing flow heterogeneity at higher spatial resolution (2, 27). Individual LV samples displayed not only threefold differences in flow, but also similar differences in local glucose and fatty acid uptake and energy turnover (10) as well as substantial differences in local gene and protein expression (17). This supported the validity of the microsphere flow measurements in this spatial domain, and prompted the question for the smallest unit displaying a rather homogenous perfusion ("microvascular unit"). The results of the present study strongly suggest that well below a spatial scale of 50 mg (µl) microspheres will not be able to answer this question. In addition, while a fit to the observed relation between the sample size and flow heterogeneity appears to open up a seductive possibility to extrapolate the heterogeneity to smaller scales, it is most likely that this will overestimate the spatial heterogeneity present. When the spatial variability of perfusion with a molecular flow marker at a resolution <1 mm2 in the rabbit heart is studied, only a very small flow heterogeneity (0.06 conduction velocity) was observed (18), much smaller than expected on the basis of microsphere measurements.
Limitations.
For different reasons, the imaging techniques employed did not allow a precise measure of the real microsphere density in the tissue examined. When epifluorescence microscopy is used (see Fig. 1), the distance of an individual microsphere from the slice surface was difficult to estimate and varied between 0 and 300 µm, influenced by the microsphere color and the optical properties of the tissue. No attempt was made to define a threshold necessary for counting microspheres. When the deposition pattern is analyzed with the use of the cryomicrotome imaging system (see Fig. 2), the resolution obtained in imaging the whole canine myocardial slice at the given pixel size exceeded the size of the individual microsphere. Consequently, two microspheres of the identical color being very close together will have been assigned a common microsphere localization. This does explain why the nearest neighbor of the same color was never observed at the distance range from 0 to 30 µm (data not shown) but could be easily seen when comparing two different colors (see Fig. 3). In fact, at the given resolution microsphere clustering even at a distance from 30–90 µm or even up to 120 µm for one color will have been systematically underestimated because under these circumstances it may not be the nearest neighbor, but rather the second- or third-nearest neighbor that is included in these statistics. Finally, when imaging iron oxide microspheres both in vivo and in vitro (see Fig. 6), both the size of the voxels imaged (2 µl and 125 nl) and the threshold observed in preliminary experiments made it impossible to reliably estimate the number of microspheres per voxel. In addition, because the average microsphere density had to exceed a certain threshold before the image intensity was affected, image areas of normal signal intensity cannot be expected to be devoid of microspheres.
In conclusion, the present study describes in detail the deposition pattern of microspheres. The unexpected findings of sequentially applied microspheres being extracted by one small arteriolar vessel of preferential channeling of microspheres to temporally stable cluster locations and of large microsphere-free voids indicate that at a high spatial resolution, the flow and deposition pattern of microspheres is not only governed by the spatial distribution of blood flow and the low probability of a given particle to end up in one of many precapillary vessels, but will also be influenced by the anatomic features of smaller segments of the arterial vascular tree. This has profound consequences for the spatial scale where microsphere can be a valid means to test the local perfusion. Previous studies supported the validity of microsphere flow determinations down to 200 mg in larger animals (3, 16) and 50 mg in smaller animals (4). The present data indicate that at a spatial scale <2 mg (µl) local flow may not be the dominant force governing microsphere deposition. Clearly, future studies are needed relating local flow to local microsphere deposition at the intermediate scales. To this end, the combination of MRI-based flow measurements (e.g., using Gd) and the application of the iron oxide microspheres introduced in this study may offer a promising avenue.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Stoll, M. Quentin, A. Molojavyi, V. Thamer, and U. K.M. Decking Spatial heterogeneity of myocardial perfusion predicts local potassium channel expression and action potential duration Cardiovasc Res, February 1, 2008; 77(3): 489 - 496. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Marxen, J. G. Sled, L. X. Yu, C. Paget, and R. M. Henkelman Comparing microsphere deposition and flow modeling in 3D vascular trees Am J Physiol Heart Circ Physiol, November 1, 2006; 291(5): H2136 - H2141. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Marsh and D. J. Ellerby Partitioning locomotor energy use among and within muscles Muscle blood flow as a measure of muscle oxygen consumption J. Exp. Biol., July 1, 2006; 209(13): 2385 - 2394. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. H. Aletras, G. S. Tilak, A. Natanzon, L.-Y. Hsu, F. M. Gonzalez, R. F. Hoyt Jr, and A. E. Arai Retrospective Determination of the Area at Risk for Reperfused Acute Myocardial Infarction With T2-Weighted Cardiac Magnetic Resonance Imaging: Histopathological and Displacement Encoding With Stimulated Echoes (DENSE) Functional Validations Circulation, April 18, 2006; 113(15): 1865 - 1870. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Fuster Symposium Presentations J. Am. Coll. Cardiol., October 4, 2005; 46(7_Suppl_A): 5A - 70A. [Full Text] [PDF]
|
|
|
|
|
|
K. W. Hewett, L. W. Norman, D. Sedmera, R. J. Barker, C. Justus, J. Zhang, S. W. Kubalak, and R. G. Gourdie Knockout of the neural and heart expressed gene HF-1b results in apical deficits of ventricular structure and activation Cardiovasc Res, August 15, 2005; 67(3): 548 - 560. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Matsumoto, T. Asano, K. Mano, H. Tachibana, M. Todoh, M. Tanaka, and F. Kajiya Regional myocardial perfusion under exchange transfusion with liposomal hemoglobin: in vivo and in vitro studies using rat hearts Am J Physiol Heart Circ Physiol, April 1, 2005; 288(4): H1909 - H1914. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
