An Iron-Binding Exochelin Prevents Restenosis Due to Coronary Artery Balloon Injury in a Porcine Model
Background Vascular smooth muscle cell (VSMC) proliferation is a critical factor in the neointima formation that causes restenosis after coronary angioplasty (PTCA). Desferri-exochelin 772SM (D-EXO), a highly diffusible, lipophilic iron chelator secreted by Mycobacterium tuberculosis, inhibits proliferation of VSMCs in culture. We hypothesized that treatment with D-EXO would inhibit neointima formation in balloon-injured vessels in vivo.
Methods and Results We subjected 24 pigs to overstretch coronary artery injury with standard PTCA balloons and then administered intramural injections of either D-EXO (n=14) or vehicle (n=10) through an Infiltrator catheter. Treatments were randomized, and the investigators were blinded with regard to treatment group until data analysis was completed. One month later, we euthanized the pigs, excised the injured coronary segments, made multiple sections of each segment, and identified the site of maximal neointima formation. An injury score based on the degree of disruption of the internal or external elastic lamina or media was assigned. D-EXO reduced stenosis index (neointima area divided by the area within the internal elastic lamina), adjusted for injury score, by 47%. Neointima thickness was also reduced.
Conclusions D-EXO, injected intramurally, substantially inhibited formation of neointima in a porcine vascular injury model.
Received May 15, 2001; revision received July 26, 2001; accepted July 31, 2001.
Although coronary angioplasty is a widely used treatment for obstructive coronary artery lesions, the success of the procedure has been limited by a high incidence of restenosis. A recent study estimated the incidence of physiologically significant restenosis to be 32% after standard balloon angioplasty and 18% after intracoronary stenting.1 Although numerous experimental approaches for preventing restenosis have been tested with varying degrees of success, none have been shown to offer long-term effectiveness.
Proliferation and migration of vascular smooth muscle cells (VSMCs) to form a neointima within the vascular lumen is a major component of coronary artery restenosis.2,3 Although VSMCs are normally quiescent, injury due to angioplasty may cause them to enter into the cell cycle and proliferate. One of the factors necessary for cell proliferation is iron. We have reported that a novel iron chelator, desferri-exochelin 772SM (D-EXO), reversibly arrests the growth of cultured VSMCs without toxicity.4 D-EXO specifically prevents progression at 2 points in the cell cycle: from G1 to S phases and from S to G2M phases.4 Because of this unusual dual action of D-EXO at separate points in the cell cycle, we postulated that this agent might be effective as a treatment for preventing restenosis.
To assess this possibility, we studied intramural catheter administration of D-EXO in a porcine model of coronary artery restenosis. Using an investigator-blinded, randomized treatment protocol, we observed the effect of D-EXO on growth of neointima 1 month after balloon injury.
Synthetic D-EXO, >98% pure and <2% iron-saturated, was provided in powder form (Keystone Biomedical, Inc). Immediately before use, the D-EXO was dissolved at a concentration of 0.83 mg/mL in 0.09% saline by brief heating to 75°C.
The experiments were performed according to the guidelines of the National Institutes of Health Guide for Care and Use of Laboratory Animals. The University of Colorado Health Sciences Center and Colorado State University Animal Research Committees approved the experimental protocol. Thirty-eight juvenile, domestic swine, weighing 16 to 35 kg, were used.
The night before the procedure, each pig received aspirin 325 mg orally. The pigs were anesthetized with acetylpromazine (5 mg/kg IM) and ketamine (35 mg/kg IM), intubated, and mechanically ventilated with isoflurane (1% to 2%) and oxygen. Reflexes and ECG and ventilatory status were continuously monitored. After a surgical cutdown, the right or left femoral artery was cannulated. The pigs were then given heparin intra-arterially (150 U/kg every hour), bretylium (5 mg/kg every hour), and cefazolin (1 g bolus). To maximally dilate the vessel and avoid spasm, nitroglycerin (200 μg) was injected into the lumen of the left coronary artery. Coronary arteriography was then performed by injection of Hypaque 76%, with images recorded in the 45° left anterior oblique and 30° right anterior oblique views. An angiographic landmark in the left anterior descending or left circumflex coronary artery, near an obtuse marginal or diagonal branch, was selected as the site for injury. With the catheter diameter used as a standard, vessel diameter was estimated at this site, and an appropriate balloon angioplasty catheter, oversized by ≈0.5 mm, was inserted.5 Either Cordis Predator or Avenger angioplasty balloons, each 20 mm long, were used. Balloons were inflated 3 times for 30 seconds, with inflations separated by 2 minutes of reperfusion time.
The Infiltrator catheter for intramural injection is a triple-lumen balloon catheter in which a drug delivery port is connected to injection needles on the surface of the balloon.6 When the balloon is deflated, the needles are not in contact with the vessel wall. When the balloon is inflated, however, the injection needles penetrate the internal elastic lamina (IEL) and permit injection of drug into the media. The catheter diameter sizes used (2.5, 3.0, or 3.5 mm) were selected to match the vessel diameter at the injury site. In previous studies, injections with this catheter in normal pig coronary arteries caused minimal or no disruption of the IEL or media, minimal or no smooth muscle proliferation, and no luminal encroachment.6 The catheter has been used in human subjects with coronary artery disease.7
The Infiltrator catheter balloon was inflated to 2 atm pressure, and an intramural injection of 0.6 mL (0.5 mg) of D-EXO (drug group) or an equal volume of 0.09% saline (control group) was administered. Treatments were randomized, and the investigators were blinded as to treatment. The first 18 pigs were injected once in the center of the injured region. It was subsequently recognized, however, that the injured coronary artery segments were 20 mm long and the region receiving the injection only 15 mm long. Therefore, in all subsequent pigs, 2 injections of 0.5 mg were made in adjacent, but not overlapping, sites. After the injections, repeat angiography was performed, the guiding catheter and sheath were removed, the femoral artery was repaired, and the wound was closed. Nitroglycerin paste was applied topically. Analgesics were administered orally as needed. The pigs were fed a normal chow diet and received 325 mg/d aspirin PO.
Euthanasia and Perfusion Fixation of the Heart
At 32±3 days after the procedure, the pigs were anesthetized, heparinized, and subjected to repeat coronary angiography. A lethal dose of barbiturate and potassium was administered, after which the heart was rapidly excised. The coronary arteries were perfusion-fixed with buffered 10% formaldehyde for 15 minutes at 110 mm Hg driving pressure. The hearts were stored in formaldehyde until coronary segments were removed for histological analysis.
Histology and Morphometry
The target regions of the coronary arteries were identified with the help of the angiographic images. The injured segment was cross-sectioned at 3-mm intervals and submitted in toto for histological evaluation. Cross sections (4 μm thick) were cut every 200 μm and stained with Verhoeff–van Gieson’s elastin stain. The presence or absence of IEL and external elastic lamina disruption, neointima formation, medial dissection, intraluminal thrombus, intramural hemorrhage, and inflammatory cell infiltrate was noted. Pigs with an intact IEL or occlusive thrombus were excluded from analysis. The section with the most severe luminal narrowing was selected for morphometric quantification. An ordinal injury score, modified from that of Schwartz et al,8 was assigned (Table 1).
Before performance of the data analysis, it was decided to exclude results in pigs that either had no evidence of IEL disruption (injury score 0) or pigs that had an organized luminal clot. All other pigs that completed the protocol were analyzed.
By use of a planimetry program (Image-Pro Plus), the borders of the external elastic lamina, IEL, and vessel lumen were traced on a digitizing board. The neointima area of each section was determined by subtracting the area within the residual lumen from the area within the IEL. Neointima thickness was the maximum thickness of the neointima measured from the adjacent IEL. The morphometric parameters were assessed separately by 2 authors (T.J.B. and E.A.R.); both were blinded as to treatment group. Interobserver variability was <5%.
Data Analysis and Statistics
Maximal neointima thickness was measured directly. Because for the same degree of stenosis, more neointima formation would be expected in larger arteries than in smaller arteries, “normalized neointima thickness” was also obtained by dividing the measured thickness by the area within the IEL. “Stenosis index” was defined as the neointima area divided by the area within the IEL.8 The outcome variables analyzed were neointima thickness (mm), normalized neointima thickness (1/mm), and stenosis index (a unitless proportion ranging from 0 to 1).
Neointima thickness, normalized neointima thickness, and stenosis index were plotted as a function of the injury score to adjust for any differences in degree of injury between control and drug groups. The slopes of linear regression lines for each group, forced through the origin, were compared. Mean values of thickness, normalized thickness, and stenosis index, adjusted for injury score, in the control and drug groups were compared by t test. These adjusted means were calculated by ANCOVA for each outcome variable, with injury score as a covariate.9,10 The probability values calculated from comparison of regression slopes and from comparison of adjusted means are necessarily equal. Data are expressed as mean±SEM. A 2-sided value of P<0.05 was considered to be statistically significant.
As noted above, the investigators performing either the animal procedures or the postmortem data analysis were unaware of whether a pig received D-EXO or vehicle. There were 6 “periprocedural” deaths the day of the procedure and 4 late deaths that occurred 1 to 3 days after the procedure. Four of the periprocedural deaths were due to ventricular fibrillation during balloon injury, and 2 were due to inadequate oxygenation. Three of the pigs that died in the periprocedural period had not received an intramural injection, 1 had received vehicle, and 2 had received D-EXO. In the late death group, 2 pigs had received vehicle and 2 had received D-EXO. In all 4 late deaths, pneumonitis and pericardial thickening were observed at necropsy.
Of the 28 pigs that completed the protocol, 4 were excluded from analysis because of failure to meet preset criteria: 3 because no IEL disruption occurred (1 control and 2 drug groups) and 1 because a total occlusion with organized clot was present (drug group). When the IEL remains intact after a procedure, there is usually little or no neointimal formation regardless of whether a treatment is applied. When an organized clot is present, the pathophysiology of the neointimal response to injury is altered. Because either of these circumstances makes interpretation of responses to a treatment difficult, these 4 pigs were excluded to avoid bias.
The remaining 24 pigs were analyzed. The individual data are shown in Table 2. The 10 control pigs received vehicle only and had injury scores of 0.5 to 3.0 (mean 1.9±0.2). The 14 drug pigs had injury scores of 1.5 to 3.0 (mean 2.4±0.2). The trend toward lower injury scores in the control group was not statistically significant. Pretreatment lumen size was estimated as the area circumscribed by the IEL. Although it is impossible to exclude some overestimation of lumen size in some pigs because of remodeling, there was no statistically significant difference between the control and drug groups. In addition, we calculated the ratio of the cross-sectional area of the balloon used in each pig to the lumen size circumscribed by the IEL. The ratios were 3.85±0.47 for the control group (n=9) and 4.26±0.44 for the treated group (n=14) (P=0.63). Figure 1 shows coronary artery sections from 2 pigs with similar injury scores: 1 control and 1 treated with drug.
Plots of individual values for neointima thickness, normalized neointima thickness, and stenosis index, each as a function of injury score, are shown in Figure 2. Least-squares linear approximations were calculated separately for control and drug-treated pigs for each parameter. Within the control group, r values for the correlation coefficient when the intercept was forced through zero were 0.96, 0.95, and 0.96 for neointima thickness, normalized neointima thickness, and stenosis index, respectively. Corresponding r values in the drug group were 0.81, 0.89, and 0.86. There were statistically significant reductions in the slopes of the linear regression lines in the drug group relative to the control group: P=0.03 for neointima thickness, P=0.001 for normalized neointima thickness, and P=0.004 for stenosis index.
Mean values of neointima thickness, normalized neointima thickness, and stenosis index, adjusted for injury score, are shown in Figure 3. Each adjusted mean is evaluated at the overall mean injury score of 2.15. Because of nonsignificant between-group differences in injury score, this adjustment slightly increases the mean for the control group and decreases the mean for the drug group.
The measured neointima thickness was 0.58±0.11 in the control group and 0.44±0.10 in the drug group (P=0.03). The normalized neointima thickness was 0.29±0.03 in the control group and 0.15±0.02 in the drug group (P=0.001). The stenosis index was 0.49±0.06 in the control group and 0.26±0.06 in the drug group (P=0.004). Therefore, treatment with D-EXO reduced neointima thickness by 24%, normalized neointima thickness by 48%, and stenosis index by 47%.
Some pigs received 1 injection and some 2, as described in the Methods. Further ANCOVA was performed with “number of injections” as a covariate. For normalized neointima thickness and stenosis index, there were no significant differences between 1 and 2 injections. For unnormalized neointima thickness, there was a marginally significant (P=0.04) change in this variable, with a coefficient of −0.20. On the basis of this coefficient, after control for group and injury score, the mean neointima thickness was 0.20 mm less for pigs with a single injection. This difference in measured neointima thickness, which applied to both control and drug groups, had no effect on the analysis of drug treatment effect.
There was a borderline difference in distribution of injury scores between the 2 treatment groups (P=0.06) due to low injury scores in 2 pigs (0.5 and 1.0) in the control group but no pigs with these injury scores in the drug group. To assess whether this difference influenced the conclusions, the data were reanalyzed without these 2 pigs, including all 14 pigs in the drug group and the remaining 8 pigs in the control group. There were again significant differences between the control and drug groups in neointima thickness (P<0.03), normalized neointima thickness (P<0.002), and stenosis index (P<0.004). We concluded that the 2 pigs with very low injury scores did not alter the results.
Various methods to prevent coronary artery restenosis have been proposed. Insertion of genetic material, such as antisense oligodeoxynucleotides, has been investigated in animal models.11 Clinical studies with ACE inhibitors, calcium blockers, colchicine, or HMG-CoA reductase inhibitors have been unsuccessful.12–14 Coronary irradiation resulted in early reductions in restenosis rates in recent clinical trials, but long-term outcomes are unknown.15,16
Because cell proliferation is iron-dependent,4 iron chelation is a potential means of preventing restenosis. One mechanism by which this may occur is through interruption of iron-mediated redox processes during the early inflammatory response to vascular injury. Reactive oxygen species released as a result of vascular injury can stimulate release of growth factors that induce quiescent VSMCs to transform into a synthetic phase.17 Treatment before angioplasty with probucol, an antioxidant that inhibits lipid peroxidation, has reduced restenosis in an animal model18 and in a human clinical study.19 By blocking iron-requiring oxidant reactions,4 iron chelators may have a similar effect. In addition, iron chelators inhibit an iron-requiring enzyme, ribonucleotide reductase, that is essential to replication of deoxyribonucleic acid.20
Exochelins, siderophores secreted by Mycobacterium tuberculosis,21 are high-affinity iron chelators soluble in both lipid and water. The water solubility facilitates administration in standard solutions, whereas the lipid solubility facilitates rapid entry of exochelins into cells.22 In the iron-free, or “desferri-,” form, exochelins block iron-mediated redox reactions.22 We have demonstrated that D-EXO reversibly arrests the growth of human VSMCs in cell culture by blocking progression of VSMCs from G0/G1 (quiescence) to S phase (DNA replication) and from S to G2/M phase (cell division).4 By interfering with more than 1 cell-cycle-regulatory pathway, D-EXO may inhibit cell growth more effectively than a strategy that targets individual cell cycle proteins. Because this agent has dual cell-cycle effects that reversibly block the growth of VSMCs in vitro without apparent cytotoxicity and can rapidly enter cells, we tested the ability of D-EXO to prevent restenosis in an animal model.
Accordingly, we tested the effect of intramural injection of D-EXO on overstretch balloon-induced coronary artery injury in pigs. The administration of vehicle or D-EXO was randomized and investigator-blinded both at the time of treatment and during the postmortem analysis. To correct for variability in degree of injury, we stratified measurements of neointima growth by use of an injury score. In view of the linear relationships between injury score and the measurements of neointima size in the control animals in Figure 2, it appears that degree of injury is a powerful independent determinate of the degree of neointimal proliferation. Others have also presented evidence to support this approach.8,10
Our observation that intramural administration of D-EXO immediately after vascular injury inhibits restenosis underscores the importance of early events in the initiation of the restenosis process. After balloon catheter vascular injury, platelets adhere to the vascular wall and release proteins within 30 minutes.23 Previously quiescent cells move rapidly into the mitotic cell cycle during the first day after injury, but cell cycle progression declines by 48 hours.24 External radiation the first day after injury is more effective than treatment the second day,25 an observation that is consistent with the hypothesis that events during the early hours are the primary determinants of whether physiologically significant restenosis ensues.
Because most of the critical steps in activation of VSMC proliferation are in the early hours after injury, inactivation of iron-mediated processes by D-EXO during the major vulnerable period for triggering restenosis presumably occurred. Treatment with D-EXO, however, did not entirely prevent neointima formation. This incomplete effect may have reflected clearance of D-EXO before the critical period for triggering restenosis had elapsed entirely. Alternatively, the amount of drug delivered to the injured vascular wall may have been insufficient to bind all the critical iron sites. The small delivery volume of the infiltrator catheter limited the quantity of D-EXO that could be injected.
Treatment with D-EXO approximately halved the stenosis index. Although differences in protocols prohibit direct comparisons, this degree of inhibition of the restenosis process appears to equal or exceed results of most other treatments. It is risky to extrapolate postinjury observations made in normal animal vessels to the more complex atherosclerotic lesions encountered in humans. The efficacy of treatment with D-EXO in this porcine model, however, offers encouragement that this agent may be effective in prevention of clinical restenosis. Because D-EXO reversibly inhibits cultured human VSMC growth without cell injury,4 toxicity from this agent is less likely to occur than with use of radiation or antisense technology. Either higher dosages of drug or exposure of injured tissue to drug for longer periods of time could enhance the results of treatment with D-EXO to prevent restenosis. Modifications in formulation could increase the concentration of D-EXO in the injectate. Other means of delivery, such as stents coated with D-EXO, may be feasible.
This work was supported by grant HL-55291 from the National Institutes of Health.
Drs M.A. Horwitz and L.D. Horwitz are major stockholders in Keystone Biomedical, Inc, which supplied the exochelin.
Liu MW, Roubin GS, King SBIII. Restenosis after coronary angioplasty: potential biologic determinants and role of intimal hyperplasia. Circulation. 1989; 79: 1374–1387.
Pahl PMB, Yan X-D, Hodges YK, et al. An exochelin of Mycobacterium tuberculosis reversibly arrests growth of human vascular smooth muscle cells in vitro. J Biol Chem. 2000; 275: 17821–17826.
Winer BJ. Statistical Principles in Experimental Design. 2nd ed. New York, NY: McGraw-Hill; 1971.
13. Faxon DP. Effect of high dose angiotensin-converting enzyme inhibition on restenosis: final results of the MERCATOR study, a multicenter, double-blind, placebo-controlled trial of cilazapril: the Multicenter American Research Trial With Cilazapril After Angioplasty to Prevent Transluminal Coronary Obstruction and Restenosis (MERCATOR) Study group. J Am Coll Cardiol. 1995; 25: 362–369.
Wardeh AJ, Kay IP, Sabate M, et al. β-Particle-emitting radioactive stent implantation: a safety and feasibility study. Circulation. 1999; 100: 1684–1689.
Rao GN, Berk BC. Active oxygen species stimulate vascular smooth muscle cell growth and proto-oncogene expression. Circ Res. 1992; 70: 593–599.
Schneider JE, Berk BC, Gravanis MB, et al. Probucol decreases neointimal formation in a swine model of coronary artery balloon injury: a possible role for antioxidants in restenosis. Circulation. 1993; 88: 628–637.
Nyholm S, Mann GJ, Johansson AG, et al. Role of ribonucleotide reductase in inhibition of mammalian cell growth by potent iron chelators. J Biol Chem. 1993; 268: 26200–26205.
Gobin J, Moore CH, Reeve JR Jr, et al. Iron acquisition by Mycobacterium tuberculosis: isolation and characterization of a family of iron-binding exochelins. Proc Natl Acad Sci U S A. 1995; 92: 5189–5193.
Horwitz LD, Sherman NA, Kong Y, et al. Lipophilic siderophores of Mycobacterium tuberculosis prevent cardiac reperfusion injury. Proc Natl Acad Sci U S A. 1998; 95: 5263–5268.
Goldberg ID, Stemerman MB. Vascular permeation of platelet factor 4 after endothelial injury. Science. 1980; 209: 611–612.
Majesky MW, Schwartz SM, Clowes MM, et al. Heparin regulates smooth muscle S phase entry in the injured rat carotid artery. Circ Res. 1987; 61: 296–300.
Shimotakahara S, Mayberg MR. Gamma irradiation inhibits neointimal hyperplasia in rats after arterial injury. Stroke. 1994; 25: 424–428.