Skip Navigation

Cardiovascular Research 2001 49(3):497-506; doi:10.1016/S0008-6363(00)00285-6
© 2001 by European Society of Cardiology
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Unger, E. F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Unger, E. F
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Copyright © 2001, European Society of Cardiology

Experimental evaluation of coronary collateral development

Ellis F Unger*

Division of Clinical Trial Design and Analysis, Office of Therapeutics Research and Review, Center for Biologics Evaluation and Research, U.S. Food and Drug Administration, Rockville, MD 20852-1428, USA

* Tel.: +1-301-827-5350; fax: +1-301-827-5394 ungere{at}cber.fda.gov

Received 25 September 2000; accepted 2 November 2000


   Abstract
Top
 Abstract
1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
 6 Summary
 References
 
During the last decade, there has been great interest in the potential use of biologic agents and mechanical techniques to enhance myocardial collateral development. Available experimental methods to detect the effects of interventions designed to improve collateral function include assessment of vascular cell proliferation, quantification of vessel number and size, appraisal of myocardial perfusion, and evaluation of myocardial function. The purpose of this review is to discuss the various experimental approaches for the evaluation of coronary collateral development, highlighting the relative strengths and limitations of the commonly used animal models and methods of assessment.

KEYWORDS Blood flow; Collateral circulation; Coronary circulation; Microcirculation; Regional blood flow


   1 Introduction
Top
 Abstract
 1 Introduction
2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
 6 Summary
 References
 
Based on putative mechanisms of action, pharmacologic and mechanical therapies devised to improved collateral function through enhancement of arteriogenesis would be manifested as: (i) amelioration of reversible ischemia as a result of enhanced collateral perfusion; and (ii) improvement in left ventricular function due to increased perfusion in areas of ‘hibernating’ myocardium. Experimental opportunities to detect the effects of interventions designed to improve collateral function include evaluation of vascular cell proliferation, quantification of vessel number and size, and assessment of myocardial perfusion and function. The purpose of this review is to discuss some of the current experimental approaches for evaluation of coronary collateral development, highlighting the relative strengths and limitations of the commonly used animal models and methods of assessment.


   2 Ischemia versus infarction
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
 6 Summary
 References
 
Reversible myocardial ischemia and myocardial infarction are fundamentally different pathophysiologic processes, and the distinction between them has important ramifications with regard to experimental therapeutics. Following abrupt interruption of coronary perfusion, there is a limited time-frame for myocyte survival (hours) relative to the time required for vascular development (days). In light of this temporal difference, it is unlikely that an angiogenic mechanism, per se, could importantly affect the initial events in myocardial infarction. (On the other hand, it is not inconceivable that angiogenic mechanisms could alter the healing response to myocardial infarction.) It follows that experimental models that cause acute myocardial infarction, through direct coronary artery occlusion, or through circulatory embolization with microspheres [1], inorganic mercury [2] or thrombus [3], have limited applicability to studies of collateral development and are not considered further in this review.


   3 Experimental models for assessment of coronary collateral development
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
4 Endpoints for experimental...
 5 General principles
 6 Summary
 References
 
Virtually all approaches for the study of coronary collateral development involve chronic models of intermittent or progressive coronary occlusion.

3.1 Intermittent occlusion model
Fujita et al. demonstrated that brief, repetitive total coronary occlusion was a potent stimulus for coronary collateral development [4]. This model, widely credited to Dean Franklin, is often referred to as the ‘Franklin’ model. A coronary artery is subjected to repetitive proximal occlusions, through brief, periodic inflation of a chronically implanted hydraulic balloon occluder. Sonomicrometer crystals, embedded in the myocardium, are used to assess regional left ventricular (LV) function in the territory of the instrumented coronary artery. With initial coronary occlusions, significant LV dysfunction in the territory of the occluded artery is observed. As collateral expansion progresses, however, the degree of LV dysfunction associated with each occlusion diminishes, until the lack of LV dysfunction in response to coronary occlusion signals the development of an adequate collateral supply. Thus, the rapidity of collateral development is inversely related to the number of coronary occlusions required before the changes in regional LV function in response to balloon occlusion are abolished. The model has been both praised and criticized on the basis of its likeness to human coronary artery disease. Atherosclerotic coronary artery obstruction does not proceed gradually, but in cycles involving plaque disruption and thrombosis, followed by stabilization [5]. Proponents have noted the similarities between this process and intermittent coronary occlusion. Others have pointed out that alternations between total arterial occlusion and complete patency are artificial and have little in common with the pathophysiology of human coronary disease. Irrespective of its pathophysiologic underpinnings, a principal advantage of the Franklin model is that myocardial infarction, an important source of variability inherent in other models (see below), is not a feature of this model, because coronary occlusion is followed predictably by reperfusion with the avoidance of irreversible injury. The major disadvantages of the model are its technical complexity and labor-intensiveness, factors that may explain why it has failed to gain wide acceptance.

3.2 Ameroid constrictors
Ameroid constrictor models, initially used by Litvak in the 1950s [6] and extensively characterized by Schaper and colleagues in the 1960s [7], have afforded large numbers of experimenters the means to investigate chronic collateral structure and function, and are still used extensively today. Ameroid constrictors consist of a hygroscopic material encased in a steel sleeve. When the device is implanted around an artery, the inner hygroscopic material swells over a period of days to weeks. Because outward movement is checked by the steel sleeve, swelling is directed inward, causing arterial compression. For the most part, ameroids have been implanted in larger animals, principally dogs and pigs. Ameroids generally cause complete coronary occlusion within 2–3 weeks [7]. The common perception, that ameroids cause gradual coronary occlusion leading to myocardial ischemia with eventual collateral development, may be an oversimplification. Though ameroids cause some degree of external arterial compression as their cross-sectional diameter diminishes, they also cause mechanical trauma which can lead to endothelial damage, platelet aggregation and/or thrombus formation, and may incite a foreign body reaction with local scar formation [7]. In light of these factors, it is probably appropriate to consider the process of ameroid-induced coronary occlusion to be progressive, but not necessarily gradual. The degree to which ameroid devices cause ischemia is dependent on general coronary artery topography, the location of the constrictor along the coronary artery (i.e. proximal versus distal), its placement relative to side-branches, the extent of pre-existing collaterals, the animal's level of activity, and genetic/species differences.

A key advantage of the ameroid model is its simplicity — once implanted on a coronary artery, progressive arterial occlusion is a near-certainty. The limitations of the ameroid model can be summarized as follows: (i) hypothetically, the process of ameroid-induced coronary occlusion may be influenced by vascular tone, platelet aggregability, thrombogenicity, inflammation and fibrosis. Thus, a variety of angiogenic growth factors, as well as anti-inflammatory, antithrombotic and antiplatelet agents, have the theoretical potential to directly affect the process of ameroid-induced coronary occlusion which could, in turn, impact on the dynamics of collateral expansion. For example, basic fibroblast growth factor has been shown to be a nitric oxide-dependent vasodilator [8], and may be a coronary vasodilator as well [9], a property that might affect the process of ameroid-induced coronary occlusion in vivo; (ii) ameroid devices are associated with a variable amount of infarction (generally subendocardial) [7], as well as sudden death in a significant fraction of animals. The production of scarring, and in particular, the variation in its extent, importantly affect the overall variability of the ameroid model. This is because scarring profoundly affects the parameters that are used to assess collateral function (vascular density, cell proliferation, myocardial perfusion and function). Because the effect of the angiogenic intervention being assessed may be comparatively modest, even limited LV scarring has the potential to significantly confound the assessment of collateral function. The impact of the variability added by infarction is a requirement for larger sample sizes to overcome variability and achieve statistical significance.

With respect to pre-clinical studies in general, there are major advantages in smaller animal models, due to issues of practicality, ethics and cost. Because of the general bulkiness of ameroid constrictors, and in light of the requirement for dissection of coronary arteries to implant them, larger animals have generally been used in these studies, as noted above. Recently, however, Operschall et al. have described the use of a novel ameroid device in rabbits [10]. In this model, ameroid material simply rides on the epicardial surface, and non-absorbable suture material, threaded twice-through the ameroid and through the subjacent epicardium, is used to entrap the underlying circumflex coronary artery. Because the suture material penetrates the myocardium at a level that is deep to the circumflex, the artery is spared the trauma of direct dissection. A Doppler flow probe is used both to localize the artery, and to adjust the baseline tension in the suture. As the ameroid expands in volume, the artery is compressed by increasing traction on the suture. Using coronary cineangiography and corrosion casts, the investigators were able to demonstrate total arterial occlusion or severe stenosis in seven to eight animals after 21 days. Regional myocardial perfusion was assessed with radiolabeled microspheres at baseline, and during maximal adenosine-induced coronary vasodilatation in terminal experiments on days 7, 14 and 21. Mortality was 22%, with a number of these animals sustaining large infarctions. In surviving rabbits, infarct size was fairly substantial. Perfusion in the risk area was reduced at day 7, progressively increasing on days 14 and 21. On day 21, endocardial perfusion was approximately half that of sham-operated rabbits, whereas epicardial perfusion was essentially equivalent. The data are somewhat limited because relatively small numbers of animals were terminally studied at each time point, but the feasibility of the model was clearly demonstrated. With placement of an indwelling left atrial catheter for microsphere injections, serial perfusion assessments could be readily obtainable in future studies. The apparent disadvantage of the model appears to be its propensity for infarcts. Presumably, this could be overcome by implanting the constrictor more distally, or by using less-hygroscopic ameroid material that would swell more slowly. If such measures were ineffective, then it would be necessary to use larger numbers of animals to overcome this variability. This model, when better characterized and refined, may finally provide investigators with a small animal model of chronic, single-vessel, coronary occlusion.

Overall, the ameroid model remains a useful means to assess angiogenic interventions in vivo. The key limitations to be kept in mind include the potential effects of biologic agents on the dynamics of ameroid-induced coronary occlusion, and the confounding effects of infarction.

3.3 A variation on the Vineberg direct arterial implantation procedure
Arthur Vineberg pioneered a palliative treatment for ischemic heart disease in which an internal mammary artery, its intercostal branches transected and actively bleeding, was tunneled directly into the myocardium with the supposition that anastomoses would form between the implanted artery and the coronary circulation [11]. Though its effectiveness was never established, the operation was used clinically for a number of years prior to the advent of coronary artery bypass surgery. By ‘marrying’ the ameroid model with the Vineberg operation, a systemic artery can be brought into contact with the coronary circulation in a collateral-dependent area [12]. The implanted artery has the potential to develop a collateral circulation in communication with the myocardium, and can be instrumented with an infusion pump to administer angiogenic agents to enhance the development of systemic-to-coronary anastomoses [13]. The unique aspect of the model is its applicability to the ‘no option’ patient — the patient with severe multi-vessel disease and no patent feeder artery to serve as a collateral source. The main drawback of the model is its variability. Layered upon the usual variability of the ameroid model are a myriad of technical factors that potentially impact on the effectiveness of a Vineberg procedure, leading to substantial inherent variability in the extent of mammary to coronary anastomoses.

3.4 Species differences and selection of an animal model
The anatomy of the porcine coronary circulation is analogous to that of man, with three major coronary arteries. In contrast, the dog has essentially a two-vessel system, with a non-dominant right coronary artery supplying only the right ventricle in the vast majority of animals. Moreover, the pig has a very limited innate collateral circulation, with only sparse endocardial connections, whereas the dog is endowed with numerous, generally epicardial, innate anastomoses, that are thought to have greater potential for development than those of the pig [14]. This apparent difference has been responsible for generally greater acceptance of the pig as an animal to study angiogenic interventions, and for criticism of the dog model.

Counter to the common perception regarding the robust innate coronary collateral circulation of the dog, recent experience suggests otherwise. A number of dog studies were conducted at the National Heart, Lung, and Blood Institute, U.S. National Institutes of Health (NIH), wherein the goal was to assess the angiogenic potential of a number of biologic agents [15–21]. In these investigations, ameroid constrictors were implanted on the proximal left circumflex coronary artery (LCX) of purpose-bred hounds. Hydraulic balloon occluders were placed adjacent to the ameroid for temporary arrest of antegrade circumflex flow, enabling assessment of collateral perfusion before ameroid-induced coronary occlusion was complete. For this analysis, the earliest assessment of collateral perfusion was included for each dog. In all cases, measurements were performed in conscious dogs within 10 days of ameroid placement (generally before there was time for collateral development), during chromonar-induced, maximal coronary vasodilatation [22]. Microspheres were used to evaluate regional myocardial perfusion. The pooled data from 127 consecutive conscious dogs, obtained in ten independent experiments conducted over 7 years, provide a median ischemic zone/normal zone (IZ/NZ) perfusion ratio of 0.04 (mean 0.06, 95% confidence limits 0.05, 0.07). These data illustrate the limited nature of acute collateral perfusion in these animals, and show that variability is consistent with typical in vivo biologic data.

Also, despite the common perception regarding the propensity for dogs to develop collaterals more rapidly and extensively than the pig, the NIH data do not support this view. Fig. 1 displays regional myocardial perfusion data from the NIH dog experiments in control animals [15–21], plotted with data obtained from control pigs at other laboratories (i.e. no animals had interventions to enhance collateral function). The ordinate represents the IZ/NZ perfusion ratio, the abscissa represents time after ameroid placement. All animals had LCX ameroid constrictors; however, the studies differ in terms of whether animals were conscious or anesthetized, and with respect to the type of stress used to test maximal collateral function (i.e. vasodilators, pacing, or exercise). Perfusion was assessed using microspheres in most studies, although echocardiographic contrast was used in one. Though the data were obtained from independent studies, the comparison has utility in contrasting, in a general way, the kind of data that may be generated using the ameroid model in pigs and dogs with different methods and procedures to assess LV perfusion. The NIH canine data show a mean IZ/NZ ratio reaching ~0.4 at 38 days, rising to ~0.5 at 204 days. Similar perfusion results were observed in conscious mini-pigs by Roth et al. [23], who assessed perfusion with radiolabeled microspheres during both moderate and severe exercise. The authors reported absolute perfusion for the endocardium, mid-myocardium and epicardium. Some liberties were taken with these data for the purpose of this review, in that the means of the endocardial, mid-myocardial and epicardial perfusion values were averaged to estimate transmural perfusion, and IZ/NZ ratios were calculated from the means of these averages (the authors did not report IZ/NZ ratios, per se). The data shown are for severe exercise. In a subsequent experiment by the White and Bloor group [24], adenosine was used to induce maximal coronary vasodilatation, and radiolabeled microspheres were used to assess perfusion. The endocardial and epicardial IZ/NZ ratios were reported separately. The mean endocardial IZ/NZ ratios ranged from 0.60 to 0.65. The epicardial ratios approached unity. For the analysis in Fig. 1, the endocardial and epicardial ratios were simply averaged, and ranged from 0.77 to 0.80. Harada et al. used colored microspheres to assess collateral perfusion in conscious Yorkshire pigs during pacing [25], and observed a mean IZ/NZ perfusion ratio of 0.80. Hariawala et al. used colored microspheres to assess perfusion in anesthetized Yorkshire pigs during adenosine stress, and observed a mean IZ/NZ ratio of 0.97 [26]. Giordano et al. assessed perfusion in conscious Yorkshire pigs during pacing, using echo-contrast to assess perfusion [27]. Mean IZ/NZ ratios ranged from 0.46 to 0.57.


Figure 1
View larger version (19K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Ischemic zone to normal zone perfusion ratios (y-axis) obtained in dogs and pigs with LCX ameroids. Data as reported by various laboratories using different methods for vasodilatory stress. The x-axis refers to time after ameroid placement. Numbers in brackets refer to references in the text. See text for details.

 
From this figure, it is apparent that collateral function in dogs, assessed in the conscious state during maximal coronary vasodilatation, does not vastly exceed that of pigs (in fact, it appears to be less). It must be emphasized, however, that the apparent IZ/NZ ratio is dependent not only on collateral function, but also on the magnitude of stress. This is because collateral perfusion is generally adequate under basal conditions, at which time the IZ/NZ ratio approaches 1 (in the absence of scarring). The ratio decreases with increasing stress. Thus, although the apparent paradox between collateral function in dogs and pigs illustrated in Fig. 1 could be explained by a true inter-species difference in collateral function, a relative inability to achieve an adequate vasodilatory stress in pigs could be contributory. The extent to which these factors are operative cannot be determined. Importantly, variability in the level of vasodilatory stress leads to variability in the assessment of collateral function. Thus, every effort should be made to maximize vasodilatory stress. If this cannot be accomplished, then stress should be standardized to the greatest extent possible. The overriding goal in planning perfusion studies should be the acquisition of high quality data, capable of uncovering the functional limitations of collaterals under standardized conditions with minimal variability.


   4 Endpoints for experimental assessment of arteriogenesis
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
5 General principles
 6 Summary
 References
 
4.1 Vascular cell proliferation
Vascular cell proliferation is a key step in arteriogenesis, as elegantly demonstrated by Schaper and colleagues some three decades ago [28]. The identification of DNA synthesis in vascular cells provides a means to identify vessels in the process of remodeling. The [3H]thymidine once used to label S-phase cells has been supplanted by non-radioactive methods. Principally, these involve administration of exogenous bromodeoxyuridine (BrdU) followed by anti-BrdU immunostaining techniques [29], as well as immunostaining for proliferating cell nuclear antigen (PCNA) [30]. The latter technique can be performed on archival tissues, because pre-treatment is unnecessary. These methods are of value for distinguishing actively remodeling vessels from quiescent vasculature. (In stable vessels, labeling is practically negligible due to low vascular cell turnover.) It should be pointed out that the immunostaining methods provide data that are quantitatively different from the standard [3H]thymidine labeling index, and different from each other [31,32]. Thus, each technique has quantitative applicability within individual experiments, but comparisons between studies are probably not valid. These methods can provide quantitative data that can be expressed as labeled cells per unit area. Under appropriate circumstances, the demonstration of vascular remodeling may imply that this process is causally related to improvements in collateral function. The limitations of these techniques include: (i) data are generally obtained at only a single terminal time point; (ii) standardization of these methods can be challenging, and their sensitivity and specificity are influenced by a myriad of technical factors, including tissue fixation and embedding, protocols for denaturation (heat, pepsin digestion), and the nuances of immunostaining methodology [33]; (iii) the analyses can be confounded by abundant S-phase cells in granulation tissue in areas of myocardial infarction under repair, as well as granulation tissue in apposition to epicardial collaterals related to prior instrumentation/surgery; (iv) tissue must be viewed at relatively high magnification, such that large numbers of light-microscopic fields require analysis; and (v) in light of the latter two considerations, the techniques are susceptible to sampling errors. When a rigorous sampling paradigm is not utilized, there is the opportunity for investigator bias, and blinded assessment is particularly important. In summary, assessment of cell proliferation is principally of value to demonstrate the existence and/or location of vascular remodeling. The techniques are labor-intensive, and though quantitative data are readily obtainable, they must be interpreted cautiously. The data have the potential to be confounded by local infarction and scarring.

4.2 Vascular number and size
Collaterals may develop at both the macroscopic and microscopic levels, and their means of detection and functional ramifications differ.

4.2.1 Macroscopic vessels
Poiseuille's law dictates that vascular resistance is inversely proportional to the fourth power of radius. The implication is that small changes in collateral diameter can be responsible for extremely large changes in collateral conductance. This is consistent with clinical observations, in that patients who manifest adequate collateral function (i.e. minimal angina and preserved myocardial function despite severe obstructive coronary artery disease) generally have arteriographic evidence of collaterals. Such collaterals are, by definition, macroscopic.

Macroscopic vessels can be demonstrated using arteriography in intact animals (generally with anesthesia), or by post-mortem casting. The former can be performed on a serial basis, but is associated with small but finite morbidity and mortality. Post-mortem casting cannot be performed serially; however, when performed with radio-opaque agents, the technique yields high resolution radiographs that are amenable to three-dimensional reconstruction and scanning electron microscopy [34].

There are several inherent limitations in the quantitative assessment of macroscopic vascular density: (i) accurate quantification of vessel diameter is critically important, because of the relation between vascular diameter and resistance, as above. Specifically, a 10% uncertainty in the diameter of a vessel assessed by quantitative coronary angiography leads to a 46% uncertainty in the estimate of collateral resistance (1.14=1.46); a 20% error in diameter translates into greater than a two-fold difference in resistance; (ii) the apparent diameter of a collateral is dependent on collateral tone, downstream resistance, injection technique (volume and rapidity of injection), and (for cineangiography) distance from the detector; and (iii) frame selection and vessel orientation are critical. Because these variables affect the apparent diameter of collaterals, and because of the marked amplification of error that occurs because of the forth-order relation between collateral diameter and conductance, these techniques are primarily of value for qualitative and semi-quantitative assessment of collateralization.

4.2.2 Microscopic vessels
Methods of quantifying vascular density include simple inspection of immersion-fixed or perfusion-fixed tissue, the use of special stains to identify vascular cells, and low-power electron microscopy. Direct counting of immersion-fixed tissue is notoriously inaccurate and should not be used for quantitative studies, largely because closed vessels are under-detected. To circumvent this problem, perfusion fixation, followed by plastic-embedding and thin sectioning (‘poor man's transmission electron microscopy’) has been used by some investigators [15,16]. Perfusion fixation can provide fairly uniform sections, and tends to open vessels, facilitating their detection and differentiation by size. These techniques are confounded by technical factors causing non-uniform perfusion, fixation artifacts, and collateral function itself (i.e. the ability to discriminate vessels is dependent on perfusion pressure during fixation, such that vascular density may be underestimated in areas of decreased perfusion). Alternatively, special stains have been utilized to identify vessels, independent of their status (open versus closed). These include alkaline phosphatase [35], silver methenamine [36] and Factor VIII [37]. Though these staining techniques have been widely used for many years, there are many technical nuances that make standardization difficult, and their relative accuracy, sensitivity and specificity have not been critically reviewed in the literature.

Low-power transmission electron microscopy has also been utilized for direct vessel counting [38]. A highly accurate and reproducible method in experienced hands, its primary disadvantage is the intensity of labor involved. It is also susceptible to sampling error, because only relatively small volumes of tissue can be analyzed.

In general, it is more accurate to analyze sections in non-overlapping, rectangular fields, facilitated by a computer-controlled motor-driven stage [15,16], rather than crudely counting vessels in round fields of view, expressing the results ‘per high-power field’. For the sake of general comparability, it is of value to report data in terms of vessels per mm2, as well as per fiber (capillary:fiber ratio).

The main difficulties associated with assessment of microscopic vascular density can be summarized as follows: (i) the detection of collaterals occurs against a veritable ‘sea’ of background capillaries. For example, the normal capillary density is 3000–5000 vessels/mm2 in pig, dog, rabbit and human myocardium. The number of larger conductance vessels of interest is <10/mm2 [15]. Thus, methods to assess vascular density must differentiate vessels by size; (ii) the apparent vascular density is profoundly influenced by infarction/fibrosis; (iii) depending on the technique used, discrimination of arteries from veins may be difficult; (iv) detection of vessels is dependent on tissue preparation, three-dimensional orientation, and staining techniques; (v) comparison of quantitative data between studies is probably not valid, due to technical differences. Methods for assessment of microscopic vascular density are summarized below:
Method Difficulty Accuracy
Light microscopy – immersion + +
fixation
Light microscopy – perfusion +++ +++
fixation followed by thin
sectioning, ±digital image
analysis
Light microscopy – special ++ ++
staining methods for vascular
cells
Low-power electron ++++ ++++
microscopy

In light of the limitations and difficulties of these methodologies, the routine assessment of vascular density as a screening method for angiogenic interventions seems unwarranted. On the other hand, it is only these techniques that can provide tangible evidence of a change in vascular structure at the tissue level — a change that is supportive of arteriogenesis as the mechanism of action for an angiogenic agent. Thus, there is a role for these techniques; however, meticulous attention to technical factors and adequate sampling and blinding are critical.

4.3 Blood flow and perfusion
Whereas blood flow is typically assessed using an electromagnetic or Doppler flow probe in units of volume per time, nutritive perfusion is quantified as volume per unit time per tissue mass.

4.3.1 Blood flow
Determination of blood flow is generally unsuitable for evaluation of collateral function. Flow can be measured in a primary feeder artery, or in a distal artery after it is reconstituted by collaterals. When assessed in a feeder artery, the technique overestimates collateral flow when the vessel is carrying collateral flow in addition to non-collateral flow, and underestimates collateral flow when other vessels, not assessed by the flow probe, are carrying significant collateral flow. With respect to the distal reconstituted vessel, flow enters via collaterals and exits via nutritive branches, and may be bi-directional, depending on flow probe position. Thus, measures of blood flow have considerable limitations in the assessment of collateral function, and probably should not be routinely employed.

4.3.2 Perfusion
When evaluated under appropriate conditions, tissue perfusion can provide an excellent assessment of collateral function. Generally, collateral-dependent areas exhibit reversible ischemia, manifested by adequate perfusion at rest but diminished perfusion under stress (relative to normal myocardium). Thus, the adequacy of tissue perfusion during stress provides the assessment of collateral function. Coronary vasodilators has been typically employed for this purpose; exercise and pacing have been used as well.

Assessment of perfusion generally requires a marker or tracer. Typically, microspheres (colored, fluorescent, or radioactive), echocardiographic contrast agents, MRI contrast agents, or positron emitters (positron emission tomography, PET) are used for this purpose. Microspheres have been used for decades and are particularly well-suited for use in conscious animals. After injection into the systemic arterial circulation, microspheres mix with blood and act as non-selective flow markers. However, because they are too large to traverse capillaries, they become trapped in the microcirculation, providing a ‘snapshot’ of tissue perfusion at time of injection. After a period of minutes to months, the tissues of interest are removed and numbers of microspheres are estimated per mass of tissue. The withdrawal of an arterial reference sample at a known rate during microsphere injection allows calculation of absolute perfusion [39]. Some microspheres are multiply labelled (i.e. disparate isotopes or colors). These lend themselves to serial assessment of perfusion, and/or evaluation under multiple conditions (i.e. at rest, during pacing, exercise or vasodilator stress).

During the last several decades, gamma-emitting radiolabeled microspheres have served as the ‘gold standard’ method for evaluating tissue perfusion. Their number in tissue samples is estimated by assessing the radioactivity of samples in a well-counter. Because of environmental concerns regarding radioisotopes, however, radiolabeled microspheres have been largely superseded by colored and fluorescent microspheres. The latter types of microspheres are only detectable after isolation from tissues, which is accomplished through liquefaction [40]. After separation, the numbers of colored microspheres are generally estimated by dissolving the spheres, producing a solution with absorbance proportional to microsphere number. Fluorescent microspheres are unique in that they can be directly counted using a fluorescence-activated cell sorting (FACS) device. Dual-labeled fluorescent microspheres have also been used to assess perfusion. Assessed at two wavelengths, the identity of each sphere is established by its ratio of absorbances. Non-radioactive microspheres are associated with three key disadvantages: (i) once tissues are liquefied, there is limited opportunity to reconsider or reassess the results (i.e. tissue mass, sample orientation, etc., are not recoverable); (ii) digested tissues cannot be used for morphologic analyses; and (iii) for colored microspheres, the requirements to isolate them from tissue and dissolve them constitute important sources of potential error. (The latter is not an issue for spheres counted with a FACS device.) For a general discussion of microsphere techniques, see Austin et al. [41].

Newer techniques for the assessment of tissue perfusion include contrast echocardiography and magnetic resonance imaging (MRI) modalities. These techniques may be adequate for semi-quantitative assessment of perfusion and localization of collateral-dependent areas, but they are not well-standardized or validated at the present time.

Recently, a novel histologic method has been shown to be capable of determining the three-dimensional distribution of perfusion in organs [42]. In this method, an automated cryomicrotome serially sections frozen organs at a pre-set slice thickness. The three-dimensional coordinates of up to four types of fluorescent microspheres are detected in each slice using automated filter wheels for excitation and emission. The positions of the microspheres are mathematically reconstructed into a three-dimensional structure, that can be mathematically manipulated. Regional perfusion at any given location is inversely proportional to the average distance to the nth microsphere. Though the authors did not compute absolute perfusion in their studies, they suggest that the technique could be modified to assess absolute perfusion using the reference sample method [39] by freezing blood samples and subjecting them to cryomicrotome analyses. This novel technique may be of particular value in circumstances wherein the spatial reconstruction of perfusion in an entire organ is of interest.

4.4 Left ventricular function
Expansion of a collateral network leads to enhanced myocardial perfusion, and ultimately, improved LV function. Thus, with respect to animal studies, assessment of LV function provides the most distal appraisal of arteriogenesis. LV function has been widely used in the assessment of arteriogenesis, in part because it is deemed to capture the ultimate desired effect of angiogenic interventions, and in part because of familiarity of investigators with the techniques. Similar to the assessment of LV perfusion, LV function is generally assessed at rest and during the stress of exercise, pacing, or pharmacologic stimulation, exposing the functional limitations of the collaterals.

Modalities for the experimental quantification of LV function include contrast left ventriculography, radionuclide ventriculography, echocardiography, MRI techniques and sonomicrometry. Left ventriculography and radionuclide ventriculography tend to assess global function, whereas echocardiography, MRI and sonomicrometry are capable of assessing regional LV function as well. Because models for the study of arteriogenesis tend to cause regional changes in LV function, techniques for assessment of global LV function probably lack the sensitivity necessary to adequately assess the effects of an angiogenic intervention. Thus, echocardiography, MRI and sonomicrometry are most widely used in such studies.

With any of these techniques, there are important limitations that warrant consideration: (i) LV function is highly dependent on loading conditions (pre-load, after-load), inotropic state, autonomic tone, circulating catecholamine levels, tissue oxygenation and the presence of scarring. Anesthesia has effects on heart rate, BP, inotropic state, autonomic tone, circulating catecholamine levels and tissue oxygenation. Surgery has the additional potential to induce marked fluid shifts that affect loading conditions, and to cause profound changes in autonomic tone and catecholamine levels. Thus, variability is a major concern when anesthesia, and particularly surgery, are utilized in the process of acquiring functional data; (ii) the detection of a functional abnormality at rest suggests stunning, hibernation, or infarction. If infarction is present, assessments of collateral function may be inaccurate or misleading; and (iii) in the presence of LV infarction, functional data should not be expected to be normally-distributed in the statistical sense, and statistical methods that presume normality are not valid.

Thus, measures of LV function have value in the assessment of collateral development; however, the use of anesthesia and surgery, as well as the presence of scarring, add variability that must be considered when planning these studies. The non-normality of the data must be considered in statistical analyses.


   5 General principles
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
6 Summary
 References
 
5.1 Randomization and blinding
As in any trial wherein there is random variability between subjects with respect to baseline variables, risks, and responses, implementation of randomization and blinding are extremely important. Investigator bias is possible and blinding is essential when subjective factors are involved in treatment assignment, surgical preparation, animal care, data acquisition, tissue sampling, rejection of outliers and/or data interpretation.

5.2 Anesthesia
As noted above, anesthesia has numerous effects on tissue oxygenation, autonomic tone, catecholamine levels, heart rate, BP, inotropic state, and myocardial perfusion — all with the potential to confound study results [43,44]. Whenever it is practical and humanely possible to acquire data in the conscious state, it should be considered. For example, percutaneous injection of fluorescent microspheres via a subcutaneous injection port is both practical and humane, and will permit repeated assessment of LV perfusion in the conscious state.

5.3 Serial assessment of endpoints
Serial endpoints are well-suited for a dynamic process such as collateral development. They provide greater confidence in the results, and are more robust, statistically. Generally, collateral development should be assessed on a serial basis, if possible.

5.4 Use of extensive sampling
The myocardium is a complex three-dimensional structure that is not intrinsically divided into ‘ischemic’ and ‘normal’ zones for the convenience of investigators. When samples are removed from the putative ‘ischemic’ and ‘normal’ zones, ignoring the balance of the myocardium, much information is lost. For example, the extent of the ischemic zone is not assessed, and data providing assurance that the appropriate area was actually sampled may be lacking. Assessment of LV perfusion and function around circumferential slices of myocardium provide a wealth of data. Such data should be obtained, whenever possible.

5.5 Incidental myocardial infarction
For models that assess coronary collateral development, the importance of avoiding infarction cannot be overemphasized. This is because the extent to which collateral development affects regional LV perfusion, function, vascularity and cell proliferation may be relatively small compared with the effect of infarction. This statement is true even for subendocardial and patchy infarction. For example, consider a sample comprised of a mosaic of scar and viable tissue, the latter supplied by collaterals capable of providing approximately half-normal perfusion during maximal vasodilatory stress. Under these circumstances, perfusion in scar may be ~0.1 ml/min per g, whereas perfusion in collateralized viable tissue may be ~2 ml/min per g. Thus, a 1-g sample comprised of 5% infarct by mass would have a calculated perfusion of 1.9 ml/min per g [0.05 gx0.1 ml/min per g+0.95 gx2.0 ml/min per g÷(0.05 g+0.95 g)=1.9 ml/min per g]. In contrast, an adjacent 1-g sample, supplied by essentially the same network of collaterals but comprised of 30% infarct by mass, would have a calculated perfusion of only 1.4 ml/min per g. Thus, to a very great extent, tissue composition affects observed perfusion, and great care should be taken to avoid techniques that produce infarction in these studies.


   6 Summary
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
 6 Summary
References
 
The design and analyses of studies for the experimental assessment of enhancement of coronary collateral development require careful planning, thoughtful selection of endpoints, rigorous application of randomization and blinding, and judicious interpretation of data.

Which is the best animal model to use to evaluate collateral development? Ideally, such an animal would be small, yet easily instrumented and maintained, would possess a coronary circulation and innate collateral circulation that are similar to those of man, would develop collaterals in the virtual absence of infarction, and would consistently exhibit regional abnormalities in LV perfusion and function, easily assessed, in response to simple interventions. Of course, such an animal model does not exist. In essence, therefore, the selection of an animal model is predicated on compromise — there is no ideal choice. In many cases, pragmatic factors should be considered ahead of theoretical considerations. For example, the coronary anatomy and innate collateral circulation of a pig may be more ‘human-like’ than those of some other species, making the pig a desirable model for study. However, if it is only possible for a particular laboratory to obtain data in pigs during anesthesia, and if regional LV perfusion and function data, obtained under anesthesia, are marginal in quality, then it might be advisable to obtain higher quality data in another species (the theoretical advantage regarding the innate collateral circulation of pigs notwithstanding). A similar argument applies regarding the best general endpoint to use for the assessment of collateral development. Often, this boils down a question of LV perfusion versus function. If a laboratory has experience in treadmill exercise in pigs, as well as the ability to assess LV function with sonomicrometry, then assessment of LV function during conscious exercise may be the best option. Conversely, for laboratories that do not routinely exercise large animals and must instead rely on pharmacologic or pacing stress, and for laboratories limited to LV functional assessment using transthoracic echocardiography, LV perfusion might be a better endpoint. Thus, the quality of data obtainable should be an important factor in designing studies of therapeutics for coronary arteriogenesis. Such concerns may override theoretical issues regarding a model's pathophysiologic similarity or dissimilarity to human coronary artery disease.

Time for primary review 11 days.


   Notes
 
* The opinions expressed are those of the author, and not necessarily those of the U.S. Food and Drug Administration.


   References
Top
 Abstract
 1 Introduction
 2 Ischemia versus infarction
 3 Experimental models for...
 4 Endpoints for experimental...
 5 General principles
 6 Summary
 References
 

  1. Battler A, Scheinowitz M, Bor A, et al. Intracoronary injection of basic fibroblast growth factor enhances angiogenesis in infarcted swine myocardium. J Am Coll Cardiol (1993) 22:2001–2006.[Abstract]
  2. Uchida Y, Yanagisawa-Miwa A, Nakamura F, et al. Angiogenic therapy of acute myocardial infarction by intrapericardial injection of basic fibroblast growth factor and heparin sulfate: an experimental study. Am Heart J (1995) 130:1182–1188.[CrossRef][ISI][Medline]
  3. Yanagisawa-Miwa A, Uchida Y, Nakamura F, et al. Salvage of infarcted myocardium by angiogenic action of basic fibroblast growth factor. Science (1992) 257:1401–1403.[Abstract/Free Full Text]
  4. Fujita M, Mikuniya A, Takahashi M, et al. Acceleration of coronary collateral development by heparin in conscious dogs. Jpn Circ J (1987) 51:395–402.[Medline]
  5. Fuster V, Stein B, Ambrose J.A, Badimon L, Badimon J.J, Chesebro J.H. Atherosclerotic plaque rupture and thrombosis. Evolving concepts. Circulation (1990) 82(3 Suppl):II47–II59.[Medline]
  6. Litvak J, Siderides L.E, Vineberg A.M. Experimental production of coronary artery insufficiency and occlusion. Am Heart J (1957) 53:505–518.[CrossRef][ISI][Medline]
  7. Schaper W. The collateral circulation of the heart. (1971) New York: Elsevier.
  8. Cuevas P, Carceller F, Ortega S, Zazo M, Nieto I, Giménez-Gallego G. Hypotensive activity of fibroblast growth factor. Science (1991) 254:1208–1210.[Abstract/Free Full Text]
  9. Unger E.F, Goncalves L, Epstein S.E, et al. Effects of a single intracoronary injection of basic fibroblast growth factor in stable angina pectoris. Am J Cardiol (2000) 85:1414–1419.[CrossRef][ISI][Medline]
  10. Operschall C, Falivene L, Clozel J.P, Roux S. A new model of chronic cardiac ischemia in rabbits. J Appl Physiol (2000) 88:1438–1445.[Abstract/Free Full Text]
  11. Vineberg A.M. The vineberg operation. 1. Revascularization of the heart. J Am Med Assoc (1966) 195(Suppl):43–47c.
  12. Unger E.F, Sheffield C.D, Epstein S.E. Creation of anastomoses between an extracardiac artery and the coronary circulation: proof that myocardial angiogenesis occurs and can provide nutritional blood flow to the myocardium. Circulation (1990) 82:1449–1466.[Abstract/Free Full Text]
  13. Unger E.F, Sheffield C.D, Epstein S.E. Heparin promotes the formation of extracardiac to coronary anastomoses in a canine model. Am J Physiol (1991) 260:H1625–H1634.[ISI][Medline]
  14. Cohen M.V. Coronary collaterals. clinical and experimental observations. (1985) Mount Kisco, NY: Futura.
  15. Unger E.F, Banai S, Shou M, et al. Basic fibroblast growth factor enhances myocardial collateral flow in a canine model. Am J Physiol (1994) 266:H1588–H1595.[ISI][Medline]
  16. Banai S, Jaklitsch M.T, Shou M, et al. Angiogenic-induced enhancement of collateral blood flow to ischemic myocardium by vascular endothelial growth factor in dogs. Circulation (1994) 89:2183–2189.[Abstract/Free Full Text]
  17. Lazarous D.F, Scheinowitz M, Shou M, et al. Effects of chronic systemic administration of basic fibroblast growth factor on collateral development in the canine heart. Circulation (1995) 91:145–153.[Abstract/Free Full Text]
  18. Lazarous D.F, Shou M, Scheinowitz M, et al. Comparative effects of basic fibroblast growth factor and vascular endothelial growth factor on coronary collateral development and the arterial response to injury in a canine model. Circulation (1996) 94:1074–1082.[Abstract/Free Full Text]
  19. Shou M, Thirumurti V, Rajanayagam M.A.S, et al. Effect of basic fibroblast growth factor on myocardial angiogenesis in dogs with mature collaterals. J Am Coll Cardiol (1997) 29:1102–1106.[Abstract]
  20. Lazarous D.F, Shou M, Stiber J.A, Thirumurti V, Hodge E, Unger E.F. Adenoviral-mediated gene therapy induces sustained pericardial vascular endothelial growth factor expression in dogs: effect on myocardial angiogenesis. Cardiovasc Res (1999) 44:294–302.[Abstract/Free Full Text]
  21. Rajanayagam M.A.S, Shou M, Thirumurti V, et al. Intracoronary basic fibroblast growth factor enhances myocardial collateral perfusion in dogs. J Am Coll Cardiol (2000) 35:519–526.[Abstract/Free Full Text]
  22. Bretschneider H.J, Eberlein H.J, Kabus H.M, Nelle G, Reichmann W. Die Wirkungen von 3-(β-Diäthylamino-äthyl)-4-methyl-7-carbäthoxy-methoxy-2-oxo-(1,2-chromen), einer selektiv coronarerweiternden Substanz, auf Kreislauf. Atmung Stoffwechsel Arzneim Forsch (1963) 13:255–268.
  23. Roth D.M, Maruoka Y, Rogers J, White F.C, Longhurst J.C, Bloor C.M. Development of coronary collateral circulation in left circumflex ameroid-occluded swine myocardium. Am J Physiol (1987) 253:H1279–H1288.[ISI][Medline]
  24. White F.C, Carroll S.M, Magnet A, Bloor C.M. Coronary collateral development in swine after coronary artery occlusion. Circ Res (1992) 71:1490–1500.[Abstract/Free Full Text]
  25. Harada K, Grossman W, Friedman M, et al. Basic fibroblast growth factor improves myocardial function in chronically ischemic porcine hearts. J Clin Invest (1994) 94:623–630.[ISI][Medline]
  26. Hariawala M.D, Horowitz J.R, Esakof D, et al. VEGF improves myocardial blood flow but produces EDRF-mediated hypotension in porcine hearts. J Surg Res (1996) 63:77–82.[CrossRef][ISI][Medline]
  27. Giordano F.J, Ping P, McKirnan M.D, et al. Intracoronary gene transfer of fibroblast growth factor-5 increases blood flow and contractile function in an ischemic region of the heart. Nat Med (1996) 2:534–539.[CrossRef][ISI][Medline]
  28. Schaper W, De Brabander M, Lewi P. DNA-synthesis and mitosis in coronary collateral vessels of the dog. Circ Res (1971) 28:671–679.[Abstract/Free Full Text]
  29. deFazio A, Leary J.A, Hedley D.W, Tattersall M.H. Immunohistochemical detection of proliferating cells in vivo. J Histochem Cytochem (1987) 35:571–577.[Abstract]
  30. Hall P.A, Levison D.A, Woods A.L, et al. Proliferating cell nuclear antigen (PCNA) immunolocalization in paraffin sections: an index of cell proliferation with evidence of deregulated expression in some neoplasms. J Pathol (1990) 162:285–294.[CrossRef][ISI][Medline]
  31. Connolly K.M, Bogdanffy M.S. Evaluation of proliferating cell nuclear antigen (PCNA) as an endogenous marker of cell proliferation in rat liver: a dual-stain comparison with 5-bromo-2'-deoxyuridine. J Histochem Cytochem (1993) 41:1–6.[Abstract]
  32. Ghazvini S, Kroll S, Char D.H, Frigillana H. Comparative analysis of proliferating cell nuclear antigen, bromodeoxyuridine, and mitotic index in uveal melanoma. Invest Ophthalmol Vis Sci (1995) 36:2762–2767.[Abstract/Free Full Text]
  33. Allison R.T, Best T. p53, PCNA and Ki-67 expression in oral squamous cell carcinomas: the vagaries of fixation and microwave enhancement of immunocytochemistry. J Oral Pathol Med (1998) 27:434–440.[ISI][Medline]
  34. Kwon H.M, Sangiorgi G, Ritman E.L, et al. Adventitial vasa vasorum in balloon-injured coronary arteries: visualization and quantitation by a microscopic three-dimensional computed tomography technique. J Am Coll Cardiol (1998) 32:2072–2079.[Abstract/Free Full Text]
  35. Rosolowsky M, Weiss H.R. Quantitative morphometric determination of arteriolar and capillary perfusion within ischemic rabbit left ventricle. Microvasc Res (1986) 32:99–111.[CrossRef][ISI][Medline]
  36. Kayar S.R, Weiss H.R. Capillary recruitment and heterogeneity of perfused capillary distribution in dog myocardium. Microcirc Endothelium Lymphatics (1991) 7:77–108.[ISI][Medline]
  37. Cimini C.M, Weiss H.R. Capillary perfusion in the occluded-reperfused canine myocardium: evidence for slowed reflow. Microvasc Res (1993) 46:116–127.[CrossRef][ISI][Medline]
  38. Loud A.V, Beghi C, Olivetti G, Anversa P. Morphometry of right and left ventricular myocardium after strenuous exercise in preconditioned rats. Lab Invest (1984) 51:104–111.[ISI][Medline]
  39. Heyman M.A, Payne B.D, Hoffman J.I.E, Rudolph A.M. Blood flow measurements with radionuclide-labelled particles. Prog Cardiovasc Dis (1977) 20:55–79.[CrossRef][ISI][Medline]
  40. Hale S.L, Alker K.J, Kloner R.A. Evaluation of non-radioactive, colored microspheres for measurement of regional myocardial blood flow in dogs. Circulation (1988) 78:428–434.[Abstract/Free Full Text]
  41. Austin R.E Jr., Hauck W.W, Aldea G.S, Flynn A.E, Coggins D.L, Hoffman J.I.E. Quantitating error in blood flow measurements with radioactive microspheres. Am J Physiol (1989) 257:H280–H288.[ISI][Medline]
  42. Bernard S.L, Ewen J.R, Barlow C.H, et al. High spatial resolution measurements of organ blood flow in small laboratory animals. Am J Physiol Heart Circ Physiol (2000) 279:H2043–H2052.[Abstract/Free Full Text]
  43. Reimer K.A, Jennings R.B, Cobb F.R, et al. Animal models for protecting ischemic myocardium: results of the NHLBI Cooperative Study. Comparison of unconscious and conscious dog models. Circ Res (1985) 56:651–665.[Abstract/Free Full Text]
  44. De Hert S, Vermeyen K, Adriaensen H. Effects of anesthetics on the coronary circulation. Acta Anaesthesiol Belg (1990) 41:327–343.[Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Unger, E. F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Unger, E. F
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?