Effect of Selective or Combined Inhibition of Integrins αIIbβ3 and αvβ3 on Thrombosis and Neointima After Oversized Porcine Coronary Angioplasty
Background—Thrombosis and neointima formation limit the efficacy of coronary angioplasty (PTCA). Clinical trials have implicated the adhesion molecules integrin αIIbβ3 and integrin αvβ3 in these processes. The roles of these molecules in vascular smooth muscle cell adhesion, platelet aggregation, and the thrombotic and neointimal response to oversize porcine PTCA was investigated by use of a selective αIIbβ3 antagonist (lamifiban), a selective αvβ3 antagonist (VO514), and a combined αIIbβ3/αvβ3 antagonist (G3580).
Methods and Results—In vitro, both αvβ3 inhibitors caused dose-dependent inhibition of porcine vascular smooth muscle cell adhesion to vitronectin but not to collagen type IV, fibronectin, or laminin, whereas selective αIIbβ3 inhibition had no effect. Intravenous infusions of either αIIbβ3 inhibitor in swine profoundly inhibited ex vivo platelet aggregation to ADP, whereas selective αvβ3 inhibition had no effect. In a porcine PTCA model, intravenous infusions of the integrin antagonists were administered for 14 days after oversized balloon angioplasty injury. After PTCA, there was regional upregulation of integrin αvβ3 in the developing neointima, as assessed by immunohistochemistry. Six hours after PTCA, obstruction of lumen by thrombus was reduced significantly by αIIbβ3 inhibition compared with either control or αvβ3 inhibition (mean control, 18.7%; VO514, 18.5%; lamifiban, 6.4%; G3580, 7.9%). Twenty-eight days after PTCA, there was a significant reduction of neointima with inhibitors of either integrin (mean intima/media ratio: control, 3.08; VO514, 1.33; lamifiban, 0.97; G3580, 1.32).
Conclusions—We conclude that both integrin αIIbβ3 and integrin αvβ3 participate in neointima development after experimental angioplasty.
Neointima development is characteristic of experimental and therapeutic arterial injury1 and is an important component of restenosis after percutaneous transluminal coronary angioplasty (PTCA).2 The role of platelets in restenosis is controversial.3 Platelet-derived mitogens promote smooth muscle proliferation and migration in vitro, processes thought to be central to neointima formation,4 and antagonists of these mitogens reduce neointima formation.5 6 The integrin αIIbβ3 (platelet glycoprotein IIb/IIIa) mediates binding of activated platelets to fibrinogen,7 the “final common pathway” of aggregation. αIIbβ3 inhibitors therefore represent powerful tools with which to test the hypothesis that platelets contribute to neointima development.8 The first clinical trial of an αIIbβ3 antagonist, the EPIC trial, showed a reduction both in ischemic complications and in clinical restenosis, apparently supporting this hypothesis.9 The humanized monoclonal antibody (abciximab) used in the EPIC trial, however, binds to the integrin β3 chain, thus inhibiting both αIIbβ3 and αvβ3.10 αvβ3 is more widely distributed than αIIbβ3 and has diverse functions, including mediating migration and proliferation of vascular smooth muscle cells (VSMCs).11 12 αvβ3 expression is upregulated after arterial injury13 and thus may play a role in neointima formation. Effects on one or both of these integrins could have mediated the reduction in restenosis reported in the EPIC trial.
Animal studies of platelet inhibition or depletion on neointimal responses to arterial injury have shown conflicting results, whereas most studies with selective αvβ3 antagonists or peptides with wider integrin specificity have shown reductions in neointima development.13 14 15 16 17 These studies differ in model, specificity of agents, dose and duration of treatment, and level of platelet inhibition achieved. It is not possible to compare separate studies of antiplatelet agents and anti-αvβ3 therapies. To resolve these issues, we undertook a comparative study using a porcine PTCA model. A selective αIIbβ3 antagonist (lamifiban),18 a selective αvβ3 antagonist (VO514),19 and a combined αIIbβ3 and αvβ3 antagonist (G3580)20 were used to attempt to discern the relative contributions of these molecules to the coronary arterial response to PTCA. Additional experiments were performed to assess the effects of these compounds on porcine VSMC adhesion and to examine the expression of αvβ3 in the porcine coronary artery after PTCA.
The structures of lamifiban18 and G358020 have been described previously. Lamifiban is a peptidomimetic agent selective for αIIbβ3 (IC50 for fibrinogen binding to recombinant human αIIbβ3, 1.4 nmol/L; IC50 for vitronectin binding to recombinant human αvβ3, >1 000 000 nmol/L). G3580 is a cyclic peptide (IC50 for fibrinogen binding to αIIbβ3, 1.5 nmol/L; IC50 for vitronectin binding to recombinant human αvβ3, 8 nmol/L). VO514 is a peptidomimetic selective for αvβ3 (IC50 for fibrinogen binding to recombinant human αIIbβ3, >10 000; IC50 for vitronectin binding to recombinant human αvβ3, 2 nmol/L).21
Six-well plates (Iwaki) were coated with fibronectin 2 μg/cm2, laminin 2 μg/cm2, collagen type IV 10 μg/cm2, or vitronectin 125 ng/cm2 (Sigma) by overnight incubation at 4°C. Porcine VSMCs were derived by explantation and used as confluent cultures (passages 2 to 5) suspended in serum-free medium. Cells (1×105) in 1 mL of medium were added to each well after serial dilutions of antagonist or vehicle (10-μL volume). Plates were incubated at 37°C for 2 hours, then washed 3 times with PBS. Adherent cells were quantified by counting 10 random high-power fields (×200) and expressed as percentage of control. Experiments were carried out in triplicate.
Porcine PTCA and Intravenous Infusions
Experiments were carried out in accordance with Home Office regulations. Juvenile male Sus scrofa were anesthetized with ketamine (33 mg/kg) and enflurane (3% to 5%). A dual-lumen Hickman line (Bard) was placed in the right external jugular vein, and an 8F arterial sheath (Bard) was placed in the right common carotid artery. Animals were heparinized (200 IU/kg). At time 0, an intravenous bolus of saline, lamifiban (0.2 mg/kg), G3580 (1.5 mg/kg), or VO514 (100 μg/kg) was administered, followed by an intravenous infusion via a CADD-1 infusion pump (Sims Deltec). Doses of αIIbβ3-inhibiting agents were adjusted according to ex vivo platelet aggregation: lamifiban 0.12 to 0.21 mg · kg−1 · h−1 and G3580 0.875 to 1.5 mg · kg−1 · h−1. VO514 was infused at a constant rate (0.6 mg · kg−1 · h−1), a dose suggested from earlier experiments as likely to produce complete αvβ3 inhibition (data not shown). The target level of inhibition of platelet aggregation with lamifiban and G3580 was 80%. Quantitative coronary angiography was performed (IDIS) to identify a segment in the right coronary artery and left anterior descending coronary artery that would produce a 1.3:1-sized inflation with a standard 20-mm coronary angioplasty balloon. Each artery was dilated twice for 30 seconds at 8 atm between 30 and 60 minutes.
To assess the thrombotic response, 2 animals in each group were killed at 6 hours with intravenous pentobarbitone. Coronary arteries were dissected out, divided into 2- to 3-mm blocks, immersion-fixed in 10% formalin for 24 hours, embedded in paraffin, and sectioned. Perfusion fixing was avoided to prevent dislodgment of thrombus. To assess neointima formation, infusions continued for 14 days in the remaining animals. Animals were killed at 28 days with pentobarbitone, and hearts were removed, perfusion-fixed with 10% formalin, and sectioned as above.
Platelet-rich plasma was produced by centrifugation of citrated (0.38%) blood (11 seconds at 13 000 rpm) and aggregated in an Aggrecorder aggregometer (DIC) with 20 μmol/L ADP (Sigma) as agonist.
Sections with a breached internal elastic lamina (IEL) were analyzed. Arterial dimensions and areas were measured by a single observer blind to treatment using computerized image analysis (SeeScan). Thrombus area is expressed as percentage obstruction of lumen area (lumen area in vivo was derived from the perimeter of IEL). Neointima was assessed by intima/media area ratios, and injury score was determined by percentage breach in IEL.
Expression of αvβ3 was assessed by use of archival sections from porcine PTCA experiments conducted in the same manner. Sections were from animals killed at 1, 6, and 18 hours and 3, 7, 14, and 28 days after injury. Six injured arteries were examined from all time points except 28 days, at which point 38 injured arteries were examined. Sections were dewaxed, and nonspecific binding was blocked with 3% H2O2 and 1% dried milk/PBS. Anti-human αvβ3 monoclonal antibody 5H8 (Genentech Inc) was used to detect porcine αvβ3. Antibody specificity was validated immunocytochemically, biochemically, and functionally in tests using pig osteoclasts and other tissues (data not shown). 5H8 was applied in 1:100 dilution for 1 hour at room temperature. After application of a biotinylated secondary antibody and the Vector ABC kit, DAB solution was added and sections were counterstained with Carazzi’s hematoxylin.
Sections were graded by 2 blinded observers as − (negative), + (1% to 33% cells stained), ++ (34% to 66% positive), and +++ (>67% positive).
Plasma drug levels were taken from 2 animals per group. Whole blood was collected into EDTA (0.17%) and centrifuged at 13 000 rpm for 5 minutes to produce plasma. This was assayed for lamifiban, G3580, or VO514 by a previously described method.22
Data are expressed as mean±SEM. Statistical analysis was by 1-way ANOVA. Statistical calculations were performed with Instat (GraphPad).
Selective αIIbβ3 inhibition (by up to 100 μmol/L lamifiban) had no effect on VSMC adhesion to any of the substrates tested (data not shown). In contrast, selective αvβ3 inhibition (with VO514) or combined αvβ3 and αIIbβ3 inhibition (with G3580) dose-dependently reduced adhesion of VSMC to vitronectin (Figure 1A⇓ and 1B⇓). Neither agent affected VSMC adhesion to laminin, fibronectin, or collagen type IV in concentrations up to 100 μmol/L and 1 μmol/L for VO514 and G3580, respectively (data not shown). The IC50 for adhesion to vitronectin was ≈5 μmol/L for G3580 and ≈25 nmol/L for VO514, higher than the quoted IC50 for adhesion of recombinant human αvβ3 binding to vitronectin with these agents.21 22 This may be due to interspecies differences, the presence of uncoated nonspecific binding sites in the well, or the expression of additional vitronectin-recognizing receptors on VSMCs.
αvβ3 Expression After Balloon Injury
As assessed by semiquantitative immunohistochemistry, integrin αvβ3 was detectable at moderate levels in the endothelium, media, and adventitia of both control and injured arteries. After injury, there was an increase in αvβ3-positive cells in the developing neointima, greatest at 28 days, the latest time point assessed in the study (Table⇓ and Figure 2⇓).
In Vivo Experiments
Sixty-one animals were used. There were 4 procedural deaths (2 of the control, 1 lamifiban, and 1 G3580 groups) and 4 later deaths of unknown cause (1 G3580, 1 VO514, 2 control). Hickman line dislodgment or damage occurred in 12 animals (4 control, 3 lamifiban, and 5 G3580 animals). No deaths or dislodgment occurred in animals used to analyze thrombus formation. Control animals whose lines became dislodged remained in the trial, whereas active-treatment animals were withdrawn from analysis. Thus, data for platelet aggregation and arterial morphometry were available for 8 animals in the group to assess thrombus formation (2 per group) and 37 animals in the group to assess neointima (10 control, 9 lamifiban, 9 G3580, and 9 VO514). Only platelet aggregation data from animals subsequently analyzed for arterial morphometry are shown.
Effect of β3-Integrin Antagonists on Ex Vivo Platelet Aggregation
Both αIIbβ3 antagonists produced sustained and equivalent inhibition of ex vivo platelet aggregation to 20 μmol/L ADP, immediately after commencement (Figure 3A⇓) and during the 14-day infusions (Figure 4⇓). Selective αvβ3 antagonism did not affect platelet aggregation compared with control. The target of 80% mean inhibition of platelet aggregation was exceeded with both αIIbβ3 antagonists (mean percent baseline aggregation over 14 days: control, 113.8%; VO514, 108.7%; lamifiban, 14.9%; and G3580, 14.2%). There were no significant differences at any time point between control and VO514 treatment or lamifiban and G3580 treatment.
Effect of β3-Integrin Antagonists on Thrombus Formation
Four injured arteries per group were harvested 6 hours after the start of the infusions. There were no significant differences in injury scores between the groups (data not shown). Treatment with either αIIbβ3 antagonist significantly reduced percentage luminal obstruction by thrombus (Figure 3B⇑) from 18.7% in the control to 6.4% in the lamifiban group and 7.9% in the G3580 group (P<0.05 for both groups). The selective αvβ3 inhibitor (VO514) had no effect on thrombus formation (18.5% versus 18.7% control).
Effect of β3-Integrin Antagonists on Neointima Formation
When the remaining coronary arteries were examined at 28 days after PTCA, 9 of 58 were found to be uninjured, with no discernible site of balloon injury (4 control, 1 lamifiban, 2 G3580, and 2 VO514). Thus, 49 injured arteries were examined (12 control, 13 lamifiban, 12 G3580, and 12 VO514). There were no significant differences in the mean injury scores between groups (control 26.9%, lamifiban 22.5%, G3580 23.5%, and VO514 21.4%). There was a significant reduction in intima/media area ratios for all treatment groups compared with control: control, 3.06; lamifiban, 0.97 (P<0.01); G3580, 1.32 (P<0.05); and VO514, 1.33 (P<0.05) (Figure 5A⇓). There were no significant differences in luminal or arterial areas between the groups (Figure 5B⇓).
Mean (SEM) plasma levels of the 3 agents during the infusions were G3580, 6.69 μmol/L; lamifiban, 1.30 μmol/L; and VO514, 8.55 μmol/L. These levels, compared with the IC50 of adhesion of VSMCs to vitronectin (see above) of each agent are shown in Figure 6⇓. Plasma levels of G3580 approximated the IC50, whereas those of VO514 were far in excess of its in vitro IC50. It is likely, therefore, that the level of inhibition of αvβ3 in the G3580-treated group was significantly less than that in the VO514 group.
As was seen in the VSMC adhesion experiments, plasma levels required for platelet inhibition with lamifiban or G3580 were higher than required for inhibition of previously published data of recombinant human αIIbβ3 binding to fibrinogen (see Methods).
We describe the first comparison of the effects of antagonists of different integrins on the arterial response to injury. These data indicate that continuous infusions of antagonists of either αIIbβ3 or αvβ3, started before and continued for 2 weeks after porcine PTCA, significantly reduce neointima formation at 28 days. αIIbβ3 inhibitors had marked effects on platelet aggregation and thrombus formation 6 hours after PTCA, whereas a selective αvβ3 inhibitor showed no effect. Equally, the αvβ3 inhibitors were able to inhibit VSMC adhesion, a property not shared by the selective αIIbβ3 inhibitor, although final reductions in neointima were comparable between the 3 treatment groups.
Both αIIbβ3 inhibition and combined αIIbβ3/αvβ3 inhibition induced rapid inhibition of ex vivo platelet aggregation, translating into concomitant reductions in adherent thrombus. Measurement of ex vivo platelet aggregation may overestimate the degree of platelet inhibition, and the relevance of this measurement to in vivo platelet function is unclear. Nevertheless, in this study there was a constant level of platelet inhibition that was equivalent between the 2 agents.
Both agents significantly reduced neointima (relative reductions: lamifiban, 65.6%; G3580, 57.6%). These reductions were of similar magnitude to the reductions in early thrombus with each agent. The lack of measurable effects of lamifiban on processes other than platelet aggregation implies that a reduction in thrombus leads, directly or indirectly, to a diminished neointimal response.
Selective αvβ3 inhibition had no measurable effect on platelet aggregation, whereas it potently inhibited VSMC adhesion to vitronectin, a function mediated by αvβ3 and other integrins sharing the αv subunit. A role for αvβ3 in platelet adhesion to the damaged vessel wall has been suggested,23 although this is usually mediated by other mechanisms, such as glycoprotein Ib/IX.24 We cannot exclude an effect of αvβ3 on platelet adherence, but our data do suggest that this is unimportant to overall thrombus formation.
At 28 days, αvβ3 inhibition reduced neointima formation by 57.1%, suggesting that the demonstrated upregulation of αvβ3 in the neointima after PTCA is functionally important in the arterial response to injury; this is consistent with other studies.13 16 25 The mechanism for this is unclear, possibly reflecting inhibition of some aspect of VSMC function.
Inhibition of both αIIbβ3 and αvβ3 with a combined inhibitor reduced neointima at 28 days compared with control, but there was no evidence of an additive effect over inhibition of either integrin alone. This may be due to a number of reasons. Retrospective plasma level data indicated that αvβ3 inhibition is unlikely to have been equivalent between G3580 and VO514. This could explain the absence of a synergistic effect with G3580, which may have been acting as a pure αIIbβ3 inhibitor. Alternatively, if inhibiting thrombus formation leads directly to less neointima, then any subsequent anti-αvβ3 effect on neointima formation would be reduced, perhaps to a level that could not be detectable with the methods and numbers used in this study.
Our data support a role for both thrombus and αvβ3 in the development of neointima. Thrombus may act as a scaffold into which VSMCs migrate via αvβ3 mechanisms. Equally, volume of thrombus may be a surrogate for other biological signals arising from platelet aggregation, including cooperation of the platelet in the coagulation cascade or release of growth factors. The weak inhibitory effect of αIIbβ3 antagonists on platelet activation by agonists released at the site of injury,26 however, suggests that αIIbβ3 inhibition exerts its influence by a reduction in the actual amount of thrombus.
αvβ3 antagonism has been shown to reduce neointima in other studies,13 14 15 17 although agent specificity has varied, and effects on αIIbβ3 are often discounted.16 αvβ3 inhibition may have effects not related solely to the recruitment of VSMCs to the thrombus in the lesion. αvβ3 also has effects on proliferation25 and apoptosis.17 27 Inhibition of these processes may affect the amount of neointima formed independently of thrombus or may affect the reorganization of thrombus by VSMCs.
Some of our data correspond with clinical experience. Most obvious is the correlation between a reduction in ischemic complications and a reduction in thrombus after PTCA. Several clinical trials of αIIbβ3 antagonists with PTCA have shown a reduction in ischemic end points during the active-treatment infusions28 with the same monitoring and platelet inhibition targets as our study. Trials using small-molecule agents or a bolus of abciximab without an infusion, however, showed that treatment postponed but did not abolish ischemic complications, with a “catching up” of events after the treatment was discontinued.25 A bolus plus infusion of abciximab is able to permanently reduce ischemic events and possibly restenosis. A definite treatment duration seems to be necessary to permanently reduce ischemic complications (and presumably thrombus formation) after PTCA. It is possible that the prolonged duration of treatment in our study exceeded this duration and thus translated into meaningful reduction of neointima.
We conclude that the neointimal response to injury engendered by angioplasty is inhibited by antagonists of either integrin αIIbβ3 or αvβ3. The reduction in neointima with αIIbβ3 antagonists appears to be related to a reduction in thrombus formation at the site of injury, whereas the reduction seen with αvβ3 inhibition is independent of any effects on platelet aggregation or thrombus formation. These findings are both supportive of and supported by clinical trials and suggest novel approaches to the problem of restenosis.
This study was funded by Genentech Inc. Professor Horton is supported by the Wellcome trust.
- Received July 10, 2000.
- Revision received August 23, 2000.
- Accepted September 11, 2000.
- Copyright © 2001 by American Heart Association
Hart CE, Kraiss LW, Vergel S, et al. PDGF-β receptor blockade inhibits intimal hyperplasia in the baboon. Circulation. 1999;99:564–569.
Coller BS. Platelet GPIIb/IIIa antagonists: the first anti-integrin receptor therapeutics. J Clin Invest. 1997;100:S57–S60.
Tam SH, Sassoli PM, Jordan RE, et al. Abciximab (ReoPro, chimeric 7E3 Fab) demonstrates equivalent affinity and functional blockade of glycoprotein IIb/IIIa and αvβ3 integrins. Circulation. 1998;98:1085–1091.
Liaw L, Almeida M, Hart CE, et al. Osteopontin promotes vascular cell adhesion and spreading and is chemotactic for smooth muscle cells in vitro. Circ Res. 1994;74:214–224.
Brown SL, Lundgren CH, Nordt T, et al. Stimulation of migration of human aortic smooth muscle cells by vitronectin: implications for atherosclerosis. Cardiovasc Res. 1994;28:1815–1820.
Srivatsa SS, Fitzpatrick LA, Tsao PW, et al. Selective alpha v beta 3 integrin blockade potently limits neointimal hyperplasia and lumen stenosis following deep coronary arterial stent injury: evidence for the functional importance of integrin alpha v beta 3 and osteopontin expression during neointima formation. Cardiovasc Res. 1997;36:408–428.
Matsuno H, Stassen JM, Vermylen J, et al. Inhibition of integrin function by a cyclic RGD-containing peptide prevents neointima formation. Circulation. 1994;90:2203–2206.
Slepian MJ, Massia SP, Dehdashti B, et al. β3-Integrins rather than β1-integrins dominate integrin-matrix interactions involved in postinjury smooth muscle cell migration. Circulation. 1998;97:1818–1827.
Coleman KR, Braden GA, Willingham MC, et al. Vitaxin, a humanized monoclonal antibody to the vitronectin receptor (αvβ3), reduces neointimal hyperplasia and total vessel area after balloon injury in hypercholesterolemic rabbits. Circ Res. 1999;84:1268–1276.
Knolle J, Baron R, Breipohl G, et al. Design and synthesis of potent and selective peptidomimetic vitronectin receptor antagonists. In: Tam J, Kaumaya P, eds. Proceedings 15th American Peptide Symposium. Dordrecht, The Netherlands: Kluwer; 1997:181–182.
Samanen J, Jonak Z, Rieman D, et al. Vascular indications for integrin av antagonists. Curr Pharm Des. 1997;3:545–581.
Gawaz M, Neumann FJ, Dickfeld T, et al. Vitronectin receptor (αvβ3) mediates platelet adhesion to the luminal aspect of endothelial cells: implications for reperfusion in acute myocardial infarction. Circulation. 1997;96:1809–1818.
Stouffer GA, Hu Z, Sajid M, et al. β3 Integrins are upregulated after vascular injury and modulate thrombospondin- and thrombin-induced proliferation of cultured smooth muscle cells. Circulation. 1998;97:907–915.