| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Cardiovascular Research Institute, MedStar Research Institute, Washington Hospital Center, Washington, District of Columbia; and 2Department of Cardiovascular Diseases and Internal Medicine, Mayo Clinic and Foundation, Rochester, Minnesota
Submitted 25 May 2006 ; accepted in final form 5 December 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
stem cell; study design
In contrast, little attention has focused on the reliability of the results of animal studies on which the human trials are based. Thus, during the course of our own angiogenesis studies using cell therapy in mice, we informally noted large interanimal variability in the anatomy of the vascular tree and the clinical perception of ischemia, even though the mice being used were genetically similar. This variability, accompanied by the fact that the results of published angiogenesis animal studies (whether using cell therapy or growth factors) are usually based on very few animals (Table 1) (1, 2, 4, 5, 7, 9–12, 14), led to the present investigation. Its purpose was to define how important interanimal variability in the adequacy of preexisting collaterals might be in determining intervention outcomes and to examine some of the protocol design strategies necessary to obtain reliable results.
|
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
MSC preparation and culture. MSCs were prepared and cultured according to our standard protocol (4). In brief, mouse bone marrow was harvested by flushing tibias and femurs with culture media (DMEM 20%). The cells were then plated for 72 h. Nonadherent cells were washed off, and adherent cells were expanded for 10 to 14 days. MSCs were purified from heterogeneous cells by negative magnetic beat selection (CD34 and CD45). These CD34- and CD45-negative cells were subsequently those used for cell therapy; i.e., they were injected into the adductor and quadriceps muscle of the mice subjected to femoral artery ligation.
For the experiment devised to determine whether we could predict outcome by using the day 1 clinical stratification, we used all BALB/c mice that were being studied during the course of the present investigation. This included mice that were treated with control media, mice that were treated with MSCs, and also mice that were treated with conditioned medium derived from MSCs (MSCCM).
The method to obtain MSCCM was already described in a previous paper from our laboratory (4). In brief, 12 ml fresh DMEM 20% were added to 70–80% confluent MSCs (175-cm2 flask). After 72 h, the MSCCM was harvested and concentrated by using special filter devices (MicroconYM-10). We employed two- or eightfold concentrated MSCCM.
Perfusion imaging. We used laser-Doppler perfusion imaging (LDPI) (Moor Instruments) to record serial blood flow measurements. Imaging was performed after limb hair removal and after mice had been placed on a heating plate at 37°C to minimize temperature variation. Calculated perfusion is expressed as the ratio of the ischemic to normal foot (LDPI ratio). Animals with amputation were scanned using the "phantom foot" approach. This technique attempts to overcome the problem arising in animals with autoamputation; many of these animals develop very good flow down to the level of the amputation. Thus, if only the intact foot were used for flow measurement, the fact that flow was severely impaired in the distal segment of the limb (leading to autoamputation) would not be accounted for. To minimize this problem, the scan area was derived from the preamputation area that was identified during baseline assessment. This scan area translates, in mice with amputation, into a phantom foot.
In vivo assessment of limb function. Functional assessment of the ischemic limb was performed using a modification of a clinical standard score. A semiquantitative assessment of ambulatory impairment of the ischemic limb was performed serially (0, flexing the toes to resist traction on the tail similar to the nonoperated foot; 1, plantar flexion; 2, no dragging but no plantar flexion; and 3, dragging of foot). The observers assigning the scores and reading the laser-Doppler scans were blinded to treatment. Two observers were involved in scoring the animals—one for each animal. Whenever the scoring was in doubt, the second scorer was consulted and a final consensus scoring was agreed to.
Stratification of mice before treatment. To minimize the effect of interanimal variability on outcome, we stratified animals by closely matching clinical score of the control and the treatment group before the application of any treatment. Therefore, we developed the following score, which combined appearance and functional assessment of the ischemic foot 24 h after surgery and before injection of cells: 1, plantar flexion, mild discoloration; 2, no plantar flexion, mild discoloration; 3, no plantar flexion, moderate to severe discoloration; and 4, any necrosis of the foot. The animals were also scanned to evaluate whether flow was predictive of outcome.
Animal preparation for high-resolution microscopic computed tomography and image analysis. A subset of BALB/c and C57BL/6 mice was prepared for microscopic computed tomography (micro-CT) imaging of the entire vascular tree as described previously (13). Briefly, contrast material (Microfil, MV122, Flow Tech, Carver, MA) was injected through the arterial system, and hindlimbs were placed in formalin and embedded in wax. Hindlimbs were imaged with our bench top, high-resolution three-dimensional (3D) micro-CT imaging system with a pixel size of 11 µm (3). The resulting 3D surface shading images (Fig. 5) were displayed by using image analysis software (Analyze version 6.0; Biomedical Imaging Resource, Mayo Clinic, Rochester, MN).
|
Data and statistical analysis. All results are presented as means ± SE. Statistical significance was evaluated by using repeated measures of ANOVA. A value of P < 0.05 was considered significant. To observe the correlation between increasing stratification score and outcome, ANOVA was used, and Tukey-Kramer honestly significant difference test was applied to do comparisons between groups. Correlations between continuous factors were done if appropriate. All statistical tests were performed with JMP IN software (version 5.1 for Macintosh).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
|
These results suggested to us the following hypothesis: Stratification, using the clinical score, predicts flow outcome at the end of the experiment. Therefore, we used the stratification process in all subsequent trials to ensure, we presumed, equal distribution of the adequacy of preexisting collaterals.
This simple clinical scoring system was highly predictive of flow recovery (Figs. 2 and 3). The same correlation was seen in each of the subgroups we analyzed, whether animals were treated with MSC, MSCCM, or control medium (data not shown). The most striking correlation was observed among score 4 animals (necrosis on day 1) and a bad recovery, which, in our BALB/c model, we defined as major amputation (amputation proximal to the metatarsal line). The sensitivity and specificity values for a score of 4 to predict major amputation was 100% and 97%, respectively.
|
|
|
High-resolution micro-CT images (Fig. 5) validated our laser flow data, demonstrating that vascular density after femoral artery ligation differs markedly not only between strains [BALB/c (Fig. 5, A and B) vs. C57BL/6 (C and D)] but also between mice within a given strain (A vs. B, C vs. D). Thus Fig. 5, A and C, shows, respectively, BALB/c and C57BL/6J mouse hindlimbs with less well-developed collaterals immediately after ligation of the femoral artery (preexisting collaterals) when compared with the mice of the same strain, depicted in Fig. 5, B and D.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
It was thus understandable, when large randomized angiogenesis trials showed results disparate from the positive studies in animal models, that virtually all attention focused on patient-related factors to explain the disparities. Investigators pointed to risk factor-related impediments to collateral development present in patients, such as diabetes, aging, and hypercholesterolemia, that are not present in standard animal models. Because of preliminary observations we made during intervention studies in the mouse hindlimb ischemia model, we focused our attention in this investigation on whether large intermouse variability exists in the functionality of preexisting collaterals and, if so, whether this might lead to false conclusions regarding the efficacy of an angiogenic intervention.
The results of the present investigation confirmed our suspicion; we demonstrate that marked interanimal variability in collateral flow occurs, despite the use of genetically inbred mice (a fact that was heretofore unappreciated) and that, in the typical animal investigational studies, which employ a relatively small number of animals, this variability can play a major role in outcome. For example, we learned that our initial positive angiogenesis results using MSCs in BALB/c mice (Fig. 1) were not real treatment effects but were most likely due to the result of unequal distribution between the treated and control groups of animals with good versus poor preexisting collaterals. Thus, when a small group of mice was randomized (but not stratified) to treatment, we found improved collateral flow and functional ischemia score. However, when a second, somewhat larger, group was randomized and, in addition, was stratified on day 1 to ensure equivalence of clinically assessed ischemia (presumably therefore also ensuring equivalence of preexisting collateral flow), MSCs did not produce any beneficial effects.
The concept that the day 1 clinical stratification assessment we employed reflects the extent and functionality of preexisting collaterals is supported by the significant correlation between scoring and laser-Doppler flow index on day 1 and the outcome measurement on day 28. The somewhat weaker correlation between laser-Doppler flow and outcome in comparison with the scoring index and outcome may be a sensitivity problem, because levels of flow index measured early after ligation are very low (<13% of normal); therefore, the range of flow at this time is extremely small, making correlations between outcome and flow immediately after ligation (whatever the technique used to measure flow) problematic. Moreover, LDPI has well-known limitations due to its low penetration of tissue (8) (i.e., it may not precisely reflect flow through collaterals situated in the muscular layer; to overcome this limitation, we measured flow just in the thin foot).
Additional evidence that the day 1 clinical stratification reflects the functionality of preexisting collaterals derives from three considerations: 1) it correlates with LDPI flow on day 1 (Fig. 4); 2) it correlates strongly with collateral flow at study termination (Fig. 2); and 3) given the brief time from femoral artery ligation to assessment (1 day), all flow supporting tissue viability on day 1 almost certainly derives from preexisting rather than from newly developed collaterals. As an aside, it is obvious that, just like LDPI, all other outcome measurements (functional assessments or histological analysis) would be highly influenced by this interanimal variability in preexisting collaterals. Thus simple concordance of multiple end points does not convey proof that the results are due to treatment rather than to unequal distribution of treatment-independent variables that can alter outcome. Very strong proof of marked interanimal differences in preexisting collaterals is obtained from a totally independent method of assessing preexisting collaterals using micro-CT imaging. Figure 5 shows micro-CT images obtained immediately after femoral artery ligation from two pairs of mice: one pair of BALB/c mice and one pair of C57BL/6 mice. Since images were obtained immediately after femoral artery ligation, any collaterals observed must be preexisting. The micro-CT images dramatically demonstrate that collateral density after femoral artery ligation can differ markedly between BALB/c mice (Fig. 5, A vs. B). Figure 5 also demonstrates that these interanimal variations are not limited to one strain—similar marked differences also exist in C57BL/6 mice (C vs. D).
This independent method, confirming the large interanimal differences in preexisting collaterals in two different strains, further emphasizes the need to take into consideration such differences in planning studies comparing the effects of interventions.
Indeed, a retrospective data analysis from studies we had performed in C57BL/6 mice (n = 81) showed a large variability in flow measured immediately after femoral artery ligation and a significant correlation from these values and flow measurements on day 28. This is in contrast with BALB/c mice, where we did not find any correlation between flow immediately after ligation and flow on day 28. This difference between the strains may be due to lower flow values of the BALB/c mice after ligation, where background noise may obscure any existing correlation. Furthermore, a small preliminary study in C57BL/6 mice showed similar significant predictive values of the day 1 score as in BALB/c mice; although, due to the more robust collaterals in C57 mice (3), no animal developed a score of 4 (data not shown).
In addition, it is noteworthy that from score 1 to 4, there is a steady worsening in outcome. This shows that the expedient of only excluding mice with major autoamputation does not eliminate the problem of interanimal variability. In our experience, it is also not sufficient to rely on simple power calculations based on prior experience with the experimental preparation, because we found large differences in outcome between mice shipped at different times, even though mice were from the same vendor and even though the same experienced operator was performing exactly the same surgery.
In terms of applying these observations to future studies, we should note that the above results are based on stratification of mice 24 h after femoral artery ligation. We did observe in other studies that prediction of outcome from clinical assessments performed immediately after ligation was not successful in BALB/c mice, because no differences could be seen immediately after ligation.
In addition, we need to emphasize that the major value of the present investigation is not to develop a prototype study protocol applicable to all strains and species: rather, it is to call attention to an issue with extraordinary important implications and one that has not previously been appreciated. Thus whatever model to investigate collateral growth is used, minimal interanimal variability in preexisting collaterals cannot be assumed. On the basis of our experience, it would be wise to assume that interanimal variability does exist and that to overcome this, either animals should be stratified in relation to preexisting collateral functionality before treatment or the numbers of animals used should be increased.
Conclusion. Our results demonstrate that there is considerable interanimal variability in the functionality of preexisting collaterals, even if animals are genetically similar, a heretofore unappreciated observation. Consequently, we believe that many of the published positive results of animal angiogenesis studies, particularly those with relatively small numbers of animals in control and treatment groups, may not reflect a true treatment effect and therefore would not serve as accurate guides for designing clinical angiogenesis trials (Table 1). We should emphasize that, although we have chosen to include only those studies relating to cell therapy in Table 1 (since our primary observations were obtained during the course of investigations studying the effects on collateral development of cells or cell products), our conclusions are relevant to all investigations studying the effects of therapeutic interventions on collateral development. Our data indicate that, to increase the likelihood of obtaining reliable conclusions, either animals should be stratified in regard to functionality of preexisting collaterals before therapy initiation, larger numbers of animals should be used than are currently typical, or both strategies should be employed. It is our belief that such design features will help obtain experimental data that will provide a more reliable foundation for planning clinical studies than is currently the case.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
|
DISCLOSURES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
|
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.
* S. Zbinden and L. C. Clavijo contributed equally to this work. 
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSDISCLOSURESREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. M. Sheridan, M. J. Ferguson, M. R. Distasi, F. A. Witzmann, M. C. Dalsing, S. J. Miller, and J. L. Unthank Impact of genetic background and aging on mesenteric collateral growth capacity in Fischer 344, Brown Norway, and Fischer 344 x Brown Norway hybrid rats Am J Physiol Heart Circ Physiol, December 1, 2007; 293(6): H3498 - H3505. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
