The clogging of arteries by neointima is a hallmark of atherosclerosis and of restenosis following balloon angioplasty. The realization in the 1980s that PDGF and its receptor play a key role in the onset of neointimal formation led us to develop PDGFR kinase inhibitors as antirestenosis agents. In this review, we describe the development of these inhibitors and their implementation as antirestenosis agents by localized delivery to the site of injury.
More than 25 years ago, percutaneous transluminal coronary angioplasty (PTCA) was introduced as an effective alternative to coronary artery bypass grafting, although the latter is still practiced when PTCA cannot be performed. The deployment of endovascular stents at the time of angioplasty improves the outcome, because the stent provides a mechanical support for the treated blood vessel, expanding its diameter. This expansion diminishes the ability of the neointima to clog the blood vessel. Even so, restenosis remains a clinical problem, affecting 20% or more of patients undergoing the procedure, as compared to 30–40% of patients who undergo balloon angioplasty without stenting. Because about 2,000,000 patients a year worldwide undergo PTCA, clinicians, academic scientists, as well as the pharmaceutical industry have been seeking ways to diminish the extent of restenosis subsequent to the procedure. Furthermore, the biochemical events leading to postangioplasty restenosis appear to be similar to those that cause accelerated atherosclerosis following heart transplantation, which leads to the clogging of the blood vessels of the transplanted heart.
2. The molecular biochemistry of restenosis
Restenosis is due to the migration of vascular smooth muscle cells (SMCs) from the media into the lumen. The SMCs migrate from the media, cross through the injured endothelium, and proliferate to generate the neointima, which clogs the blood vessel. In addition, lymphocyte infiltration appears to underlie the inflammatory component of this event.
Several growth factors and cytokines are responsible for the process of restenosis, including PDGF [1,2], fibroblast growth factor (FGF; ), and to a smaller extent IGF-1, TGF-β, and EGF. These factors originate from platelets, mononuclear cells, endothelial cells, and SMCs. Although all of these factors may play a role in driving the early cellular events leading to restenosis, it is generally accepted that PDGF, acting on the PDGFRβ on the SMC in the media, is the prime culprit [1,2], with FGF also playing a significant role. PDGF (and other factors), emanating from the aggregating platelets and activated monocytes at the site of injury and from the SMC themselves, acts both as a chemoattractant and a proliferating agent and is responsible for the migration of the SMC and the formation of the neointima.
Because of the prominent role PDGF was believed to play, already in the 1980s, we reasoned that selective inhibition of PDGF receptor (PDGFR) kinase at the site of balloon angioplasty might be of significant therapeutic benefit. We therefore began to develop PDGFR kinase inhibitors for the prevention of SMC proliferation, with the aim of utilizing these inhibitors in the prevention of restenosis and of accelerated atherosclerosis in the transplanted heart.
3. Early tyrphostins directed at PDGFR
In view of the discovery that one can generate selective inhibitors against various protein tyrosine kinases (tyrphostins; for review, see Ref. ), we examined the possibility of targeting PDGFR. We wished to explore the use of such inhibitors to inhibit the proliferation of SMC and to block balloon injury-induced stenosis in an animal model. By 1991, we had generated a number of tyrphostins that were relatively specific inhibitors of PDGFR and also inhibited the proliferation of SMC ([5,6]; Fig. 1). AG 213 (RG 50864), AG 82 (RG 50858), and AG 17 (RG 50872) were very effective inhibitors of rabbit vascular smooth muscle proliferation . AG 17 inhibited the proliferation of PDGF-dependent SMC proliferation with a much lower IC50 (0.04 μM) than that required to inhibit serum-supported proliferation (IC50=1.23 μM). AG 213, on the other hand, was 10 times less potent and was much less able to discriminate between PDGF and serum . AG 17 was also a potent inhibitor of PDGF-dependent PDGFR autophosphorylation (IC50 ∼2.5 μM) and of PLCγ phosphorylation (IC50 ∼2.5 μM) and exhibited a very low IC50 (<0.04 μM) for the inhibition of PDGF-dependent c-fos expression. AG 213 was less effective and was already known to be a better EGFR kinase inhibitor ; therefore, it was not considered for further studies as a PDGFR kinase inhibitor. Another attractive feature of AG 17 was that it was not toxic up to 90-fold the IC50 values for the inhibition of PDGF-dependent proliferation of rabbit vascular SMC .
The structures of PDGFR kinase inhibitors. The figure describes the structures of PDGFR kinase inhibitors developed until 2004.
Due to these properties of AG 17, we examined its effect on neointimal hyperplasia in the rat carotid artery injury model . Periadventitial-controlled release of AG 17 was examined on rat carotid arteries injured by balloon catheter. Histomorphometric analysis of the arteries after 21 days revealed that the mean neointima to media ratios were significantly reduced following treatment by both a polymer (ethyl vinyl acetate, EVA) containing 2% AG 17 (0.79 ± 0.17) and a polymer containing 10% AG 17 (0.59 ± 0.09) in comparison to the control group (1.38 ± 0.18, n=16). Similarly, the mean percent stenosis was significantly reduced by both treatment regimens (17.38 ± 3.48 in the 2% AG 17 group and 16.41 ± 3.40 in the 10% AG 17 group) in comparison to the controls (39.09 ± 4.49). Western blot analysis of explanted arterial segments revealed enhanced tyrosine phosphorylation in injured arteries, which was essentially restored to control levels in AG 17-treated arteries.
In these early experiments, shrinkage of the media was noted in the AG 17-treated rats , hinting that AG 17 might be toxic. Yet, the encouraging effects on stenosis led us to examine this compound in the pig carotid injury model. AG 17 was formulated into nanoparticles delivered at the balloon injury site. AG 17 was indeed found to be toxic to pig arterial SMC, and its growth inhibitory effect was not reversible in the pig system . Because the histology of the pig blood vessels is closer to that of humans and is very different from that of rodents like rats and rabbits, we decided to focus our work on pigs and on human tissue specimens when available (under the Helsinki committee guidelines).
4. Indoles and quinoxalines
Within the framework of our efforts to diversify the chemical entities that block tyrosine kinase activity (for review, see Ref. ), we developed catechol-free tyrphostins. We developed a number of compounds [5,11] that showed good selectivity for PDGFR vis-à-vis EGFR, including indole tyrphostins like AG 370 (Fig. 1; ). One of the overall strategies to generate novel tyrphostins was to incorporate the nitrogen atom in the nitrile group of the benzenemalononitrile moiety into a second ring. Four classes of compounds were generated: quinolines, isoquinolines, quinazolines, and quinoxalines [11,12]. Among the four classes, the quinoxalines were found to be the most selective and the most potent vis-à-vis PDGFR and its family members, c-Kit and Flt-3 [11,12].
4.1. Bicyclic quinoxalines
A number of bicyclic quinoxalines AG 1295 and AG 1296 (Fig. 1) were found to be highly potent and selective towards the PDGFR and its family members . In vitro, AG 1295 was found to significantly inhibit rat SMC growth stimulated by PDGF BB or fetal calf serum (FCS). This antiproliferative effect of the drug was paralleled by a reversible reduction in the total phosphotyrosine level and in the degree of PDGFRβ phosphorylation [9,13]. In the rat carotid model, AG 1295 delivered from polymeric matrices resulted in a 35% reduction in neointimal formation on day 14 following balloon injury. In the absence of treatment, a significant increase in tyrosine phosphorylation in arterial tissue extracts was noted on day 3 after balloon injury; tyrosine phosphorylation essentially returned to basal levels by 14 days after the injury. AG 1295 treatment decreased tyrosine phosphorylation at both time points to below basal levels. Similarly, PDGFRβ expression is normally enhanced at 3 and 14 days following arterial injury, and this enhancement was strongly inhibited by AG 1295 treatment .
These encouraging results led us to examine in detail the activity of this class of compounds on pig and human vascular SMC. In a detailed study conducted on SMC from the pig abdominal aorta and from the human mammary artery and on a single human carotid artery explant, we established the properties of AG 1295 . In vitro, AG 1295 exerted a marked selective inhibitory effect on the activation, migration, and proliferation of porcine and human SMC, as well as on PDGF-elicited PDGFR phosphorylation. AG 1295 also inhibited SMC proliferation in explants of human atheroma. AG 1295 was shown to act as long as it was present in the growth medium, but growth was fully reversible upon agent removal, suggesting that it would be nontoxic. Treatment of porcine carotid explants and of human atheroma explants with AG 1295 resulted in a significant delay in explant growth and up to 92% inhibition in the accumulation of SMC by the end of the 24-day experiment. Unlike AG 17, which inhibited porcine SMC and vascular endothelial cells with similar potency, AG 1295 selectively inhibited growth and accumulation of SMC and had a much smaller effect on the vascular endothelial cells. Most importantly, local intravascular delivery of AG 1295-impregnated polylactic acid-based nanoparticles to the site of controlled balloon injury in the femoral artery resulted in a ∼50% reduction of the intima to media ratio compared to contralateral arteries treated with nanoparticles containing no agent. These features of AG 1295 (and AG 1296) make the compound a strong candidate for the prevention of restenosis subsequent to balloon angioplasty. Its delivery at the injury site using biodegradable nanoparticles enables its utilization by a delivery balloon that spreads the drug-eluting nanoparticles at the injury site. Nanoparticles eluting AG 1295 (or another PDGFR kinase inhibitor) can be utilized with any delivery balloon that enables the passage of the 90–160 nm particles described in our studies. This may be advantageous over drug-eluting-coated stents, because one would not have to worry about long-term effects of the coating polymer.
4.2. The tricyclic quinoxaline AGL 2043
In view of the success we had with AG 1295, we decided to move closer to the human condition of restenosis and to test improved versions of the quinoxaline using the pig heart balloon injury model. We generated tricyclic quinoxalines, in which the third ring is an imidazole ring attached to the quinoxaline ring system (Fig. 1; ). The main reason for the generation of these more hydrophilic compounds was to allow better systemic bioavailability without damaging their ability to be formulated within various polymers. These compounds, especially AGL 2033 and AGL 2043, were found to be as good as or better than AG 1295 as selective inhibitors of PDGFR, both in cell-free assays and in cellular assays . AGL 2043 shows very good biological activity when formulated in nanoparticles released from a delivery balloon, similarly to AG1295 . In parallel, we have tested the efficacy of AGL 2043 from a drug-eluting stent in the pig model. AGL 2043 was incorporated into a drug-eluting stent utilizing a 1-μm-thick biodegradable polymer coated onto the EuroCor bioflex stent. Stents coated with the biodegradable, polylactic/glycolic acid (PLGA) polymer, with or without 180 mcg AGL 2043 were implanted in the proximal LAD of 23 Sinclair minipigs to achieve a 1.1:1 stent/artery diameter ratio. The delivery of the drug from the stent to the tissue was monitored by fluorescence microscopy and by HPLC analysis. After 28 days, histomorphometric analysis showed that AGL 2043 reduced instent stenosis by 60%, as compared with pigs in which no inhibitor was included in the coated stent. We found no difference in injury and inflammation scores between the pigs that were treated with AGL 2043-impregnated stents and those that received coated stents without the inhibitor. Drug levels in blood, distal LAD tissue, kidney, lung, and liver were zero at 1 and 24 h after stent implantation . It should be noted that the extent of restenosis induced by the polymer coated stents is identical to that induced by the same stents without the coating (data not shown). As indicated above, no inflammation is induced by the stents used in the study. Indeed, these polymer-coated stents have received marketing approval in Europe.
The inability of AGL2043 to inhibit restenosis by more than 63% strongly suggests that other signaling events are involved. Indeed, there is good evidence for the involvement of FGF and its receptor [2,3]. We believe that the ideal compound would be tyroposin phosphorylation inhibitor (tyrophostin) that blocks the PDGF and FGF receptors, but not the VEGF receptor. Such an inhibitor would be expected to reduce neointimal growth by more than 63%, but not to inhibit the proliferation of the endothelial cells, which is necessary for the repair of the endothelial damage inflicted by the stented balloon. It should be noted that quinoxalines administered systemically to pigs undergoing balloon injury inhibit neointima formation . Thus, the option of treating patients undergoing balloon angioplasty by the systemic application of PDGFR kinase inhibitors remains valid. This is especially true because the treatment is limited to a short period after the procedure. Nonetheless, one cannot predict the toxicity profile of quinoxalines, and the development costs are much higher for a drug which is to be applied systemically than for a drug for localized application.
5. Comparison of AGL2043 with Rapamycin and Taxol
The Rapamycin-coated stent received FDA approval in 2002 and has been on the market ever since. Rapamycin is an inhibitor of the protein kinase mTor and leads to arrest in the G1 phase of the cell cycle. This is similar to AGL 2043, which arrests cell growth at the G0/G1 phase of the cycle. In the pig balloon injury model, Rapamycin leads to 50% inhibition of restenosis , compared to 63% inhibition by AGL2043 in our study . This is rather surprising, because mTor is downstream to both PDGFR and FGFR (Fig. 2), and one would expect that Rapamycin will have a stronger effect. Thus, it would appear that the proliferative response to endothelial injury is not limited to the mTor pathway, and that other proliferative pathways, which are independent of mTor and lie downstream to PDGFR (and other receptor tyrosine kinases), are also involved. These include the Ras–Raf–Mek–Erk pathway, the PLCγ pathway, and more (Fig. 2). Using gene transfer of dominant negative Ras, it was shown that inhibition of the Ras pathway is effective in the rat model in inhibiting injury-related neointima formation . Recently, the anticancer agent paclitaxel (; Taxol) was reported to be effective when applied in a drug-eluting stent. Taxol, which arrests the cell cycle at the G2/M phase of the cell cycle (Fig. 2), apparently has no toxic effects on the proliferating SMC. At present, the hypothesis that inhibiting restenosis is best by utilizing PDGFR kinase-directed tyrphostins remains experimentally unproven. The main reason we favor this approach is that inhibitors like Taxol or Rapamycin would affect endothelial cells and smooth muscle cells with equal efficacy, because they target identical signaling elements in both pathways. PDGFR kinase-directed tyrphostins target smooth muscle that express these receptors but are not expected to target endothelial cells that express VEGFR but not PDGFR. Thus, during screening of PDGFR kinase inhibitors, we selected those that discriminate between the two, in favor of PDFGR. This argumentation also favors PDGFR kinase inhibitors over cell cycle inhibitors and the Ras–Raf pathway inhibitors, because these signaling modules are also common between endothelial cells and smooth muscle cells. Only long-term follow-up of patients with the various antiproliferative agents formulated on the drug-eluting stents will eventually establish which agent is the most effective. Furthermore, it is unclear yet whether the utilization of systemic treatments will not reemerge with full force. Because many of the PDGFR kinase inhibitors depicted in Fig. 1 are nontoxic, it may be possible that the systemic treatment of patients for a limited period of time, as required for patients undergoing balloon angioplasty and stenting, could be extremely beneficial.
The points of intervention of antirestenosis agents.
6. PDGFR as inhibitors of allograft vasculopathy
It has been reported that subsequent to the allograft implantation of blood vessels, allograft vasculopathy , including cardiac allograft vasculopathy subsequent to heart transplantation , involves an accelerated process of atherosclerosis. In these cases, prevention is the most logical mode of treatment. In a recent study, we examined whether PDGFR kinase inhibitors, especially tyrphostin AG 1295, could prevent allograft vasculopathy. Towards this end, rat aortic allografts transplanted from dark agouti (RT1 av1) donors to Wistar–Furth (RT1 u) recipients were treated with AG 1295 by local delivery, using a matrix wrapped around the graft immediately after transplantation. The recipient rats received no other treatment. At day 80 posttransplantation, intimal thickness in AG 1295-treated grafts was reduced by 50% compared to controls treated with the matrix alone . This encouraging result suggests that PDGFR-directed tyrphostins can be considered for a variety of pathophysiological conditions in which the PDGFR plays a significant role.
7. The future
We believe that it is most logical to arrest the growth signaling pathways at the source, namely, at the primary signaling element. It is likely that the best efficacy will eventually be achieved by an agent that targets both PDGFR and FGFR. Some believe that it is also logical to examine Cdk2/Cdk4 kinase inhibitors, because these directly target the cell cycle machinery driven by upstream signaling elements. Antisense oligonucleotides against Cdc2, applied to the injured area, inhibited neointima formation . Similarly, the overexpression of the Cdk2 inhibitor p21 by means of adenoviral gene delivery was efficacious in the rat carotid artery injury model . It is likely that signaling inhibitors targeting the cell cycle, the Ras pathway, and other signaling elements important for driving the process of restenosis will also find their way onto drug-eluting stents. The theoretical difficulty with these strategies is that they target elements in the cell cycle machinery or signal transduction pathways that are identical in the proliferating smooth muscle cells one would like to target and the endothelial cells one would like to avoid targeting. Two important aspects not discussed in this review that deserve serious consideration are the inflammatory component in the process of instent restenosis and the recruitment of extravascular cells into the neointima . The inflammatory cells, as well as PDGF, can in principle contribute to the recruitment of circulating progenitor cells. The mechanism of recruitment is still unknown, but its elucidation will most probably pave the way for novel treatments of restenosis.
I would like to acknowledge Dr. Shoshana Klein for comments and for the editing; Unresto, Jerusalem, Israel, for support of the studies described in this review; and Professors Shmuel Banai, S. David Gertz and Gershon Golomb whose collaboration with me made this work possible.
This review is dedicated to the memory of Prof. Amiram Eldor, a pioneer in prevention of restenosis, who died in a plane crash in 2001.
(1997) Primate smooth muscle cell migration from aortic explants is mediated by endogenous platelet-derived growth factor and basic fibroblast growth factor acting through matrix metalloproteinases 2 and 9. Circulation 96:3555–3560.
(1995) Adenovirus-mediated over-expression of the cyclin/cyclin-dependent kinase inhibitor, p21 inhibits vascular smooth muscle cell proliferation and neointima formation in the rat carotid artery model of balloon angioplasty. J. Clin. Invest. 96:2260–2268.