The limited information regarding the long-term safety of drug-eluting stents (DES) for the treatment of ST-segment-elevation myocardial infarction (STEMI) has raised concern regarding its clinical safety, in particular with respect to the associated morbidity and mortality attributed to late stent thrombosis (LST). Observational studies in daily practice have shown that the risk of LST continues at a constant rate up to at least four years after DES implantation.1 Given that MI is one of the few clinical presentations in which percutaneous coronary intervention (PCI) has been shown to reduce the risk of death, compared to medical therapy alone, the long-term outcomes after DES for the treatment of acute MI (AMI) is of immense clinical importance.2,3
Previous Studies of Drug-eluting Stents for Acute Myocardial Infarction
Several randomised clinical trials have compared clinical outcomes for patients with AMI treated with either bare-metal stents (BMS) or DES.4–7 A meta-analysis of these trials, with limited follow-up of one to two years, demonstrated a significant reduction in re-intervention with no differences in stent thrombosis, myocardial infarction or deaths between patients treated with BMS versus DES.8 Despite favourable clinical outcomes of generally low death and re-infarction rates of 4.6 and 3.5%, respectively, it remains questionable how applicable these results are to the broader use of this technology in the STEMI population.
Although several registries of DES usage for AMI have enrolled high-risk patients, registry data are imperfect and subject to selection bias, and have demonstrated inconsistent outcomes. For example, the Massachusetts Registry reported by Mauri et al. of AMI patients receiving either BMS or DES9 shows a significant decrease in repeat revascularisation and two-year mortality compared with BMS. However, at two days after the index procedure, the adjusted mortality for STEMI patients was already significantly lower for DES. This finding reinforces the notion that selection bias can heavily influence interpretation of data even after statistical adjustment.
In another study, Steg and colleagues published data from a large multinational registry of 5,093 patients receiving BMS or DES for AMI where there was increased late mortality (i.e. from six months to two years) for DES versus BMS patients,10 although the propensity and risk-adjusted survival post-discharge was not different at six months or one year. It is tempting to speculate that the events accounting for differences in mortality were thrombotic in nature and linked to discontinuation of antiplatelet therapy, which has been shown in multiple clinical studies to be a risk factor for the development of LST. Since follow-up data were available in only approximately 50% of patients, it is impossible to draw any firm conclusions about whether there was or a causal relationship between withdrawal of antiplatelet therapy and mortality. Despite these weaknesses, the data of Steg et al. are corroborated by our own pathological findings and should at the very least raise an awareness of the possibility of serious long-term risks with the strategy of using DES for the treatment of AMI.
HORIZONS-AMI Trial
One-year results from the HORIZONS AMI trial, in which 3,006 patients with AMI receiving aspirin and clopidogrel for one year were randomised to unfractionated heparin plus IIb/IIIa inhibitor or bivalirudin and paclitaxel (PES) or BMS (weighted 3:1),11 showed no differences in thrombosis and re-infarction rates at one year although, as expected, target lesion revascularisation was significantly lower in patients receiving PES. Although these data are encouraging, long-term follow-up beyond one year is required to confirm these results. Moreover, while this trial may provide further answers regarding the safety and benefits of DES in STEMI patients, it is not powered to address the relative risk of stent thrombosis or mortality with DES compared with BMS.
A recent presentation at the i2 summit (Orlando, American College of Cardiology, 2009) reported on a sub-analysis of intravascular ultrasound (IVUS) data from the HORIZONS trial. While neointimal volume was significantly less in PES versus BMS, the incidence of malapposition in PES was as high as 28.3%, which was significantly greater than BMS (7.9%; p<0.0009) at 13-month follow-up. This suggests that despite a similar incidence of in-stent thrombosis at one year, patients with DES may be at greater risk of very late stent thrombosis based on the higher prevalence of malapposition. Consistent with this notion, a recent meta-analysis reported that the incidence of late stent malapposition is significantly greater in DES compared with BMS12 and was associated with the occurrence of LST.
Pathophysiology of Late Stent Thrombosis in Acute Myocardial Infarction
Autopsy studies of patients dying from complications of LST attributed to DES report delayed arterial healing characterised by severe suppression of smooth-muscle-cell infiltration, persistence of fibrin and poor endothelialisation as the primary pathological substrate.13,14 Recent data from our laboratory in patients dying after DES placement demonstrated delayed vessel healing at culprit sites of AMI patients (i.e. plaque rupture sites) compared with stable lesions or non-culprit sites within the same stent, emphasising the importance of plaque morphology in biological healing of DES (see Figure 1).15 Furthermore, the prevalence of LST in patients with AMI versus stable angina was also significantly higher. Although observational studies and randomised clinical trials of patients receiving DES for AMI have yielded inconsistencies regarding the safety of this practice, the pathology data supports a risk for DES-driven LST in an infarct-related artery, which may outweigh potential long-term benefits.
There are several possibilities regarding how plaque morphology might influence healing in response to DES. Since ruptured plaques account for >75% of acute coronary thrombi and these lesions typically show large areas of necrosis, there is a greater risk of strut penetration into the necrotic core. One can argue that since sirolimus and paclitaxel are highly lipophilic,16 there is a greater tendency for these agents to accumulate in lipid-rich plaques and to exhibit longer dwell times relative to stents deployed on a base of fibrous tissue. Moreover, necrotic cores are generally less cellular and proliferative compared with fibrous plaques of stable lesions therefore, neo-intimal coverage over buried struts within a necrotic core is likely to be poor. Alternatively, increased drug uptake by the thrombus as shown by Hwang et al. with PES17 may also play a role. A clinical correlation to the last findings can be found in the data of Sianos et al., who investigated the impact of thrombus burden on clinical outcome in patients treated with DES for STEMI.18 Notably, patients with a greater thrombus burden had significantly higher mortality rates, major adverse cardiac events (MACE) and in-stent thrombosis compared with those with a small thrombus burden.
Collectively, these data highlight the complexities of DES on vascular healing in the ‘real-world’ setting of STEMI and potential complications of LST with regard to plaque morphology. Considering the current inconsistencies attributed to the safety of DES for the routine treatment of AMI in terms of mortality and risk of LST, the issue is far from settled. The human pathology data clearly indicate greater delayed healing in response to DES implanted at sites of plaque rupture versus stable plaque, which can theoretically contribute to the increased risk of late thrombotic events. Clearly, there is a requirement for larger randomised controlled trials before newer technologies such as DES are implemented for currently unapproved indications such as AMI.