Patients with diabetes have abnormalities in the growth and function of arterial smooth-muscle and endothelial cells and consequently have a particularly aggressive and diffuse pattern of atherosclerosis, leading to increased cardiovascular complications and mortality compared with the non-diabetic population.1–4 However, the molecular signalling underlying the accelerated atherosclerosis in patients with diabetes has yet to be completely elucidated. In the insulin-resistant state, the phosphoinositide 3-kinase (PI 3-kinase) pathway is downregulated while the pro-mitogenic mitogen-activated protein kinase (MAPK) pathway remains intact.5,6 The imbalance between these two offsetting pathways tends to favour an increase in smooth-muscle cell (SMC) migration and proliferation,7,8 potentially increasing both background atherosclerosis and restenosis following balloon angioplasty or bare-metal stent (BMS) placement in patients with diabetes compared with patients without diabetes.9–11 In addition to increased restenosis within the treated segment, patients with diabetes are also more likely to require revascularisation proximal or distal to the treated segment itself due to accelerated atherosclerosis in the remainder of the vessel.
Drug-eluting stents (DES) reduce the rate of restenosis and clinically driven repeat revascularisation compared with BMS in general populations12,13 and in patients with diabetes.3,14 Moreover, mortality in DES-treated patients with diabetes is similar or decreased compared with BMS-treated patients with diabetes (although increased compared with patients without diabetes).4,15–17 However, it is unclear whether all DES reduce the excess neointimal hyperplasia observed in the diabetic versus the non-diabetic state to an equal degree, or do so with the same level of safety. This article will focus on the bench-to-bedside evidence in terms of the effectiveness of the two major drug classes (paclitaxel and -olimus drugs) employed in DES to reduce restenosis in patients with diabetes compared with patients without diabetes.
Basic Drug Mechanisms
The antiproliferative agents used in current commercial DES derive from two distinct categories with different mechanisms of action (see Figure 1). The -olimus (rapamycin) drugs, including sirolimus (CYPHER), everolimus (XIENCE V/PROMUS), biolimus A9 (BIOMATRIX, NOBORI) and zotarolimus (ENDEAVOR, RESOLUTE), act on mammalian target of rapamycin (mTOR), a major intermediary in the PI-3-kinase pathway.18
In the diabetic state, the PI-3-kinase pathway is downregulated while an alternative Ras/MAPK pathway remains uninhibited and becomes the dominant pathway mediating SMC proliferation.6,19 By contrast, paclitaxel (TAXUS) acts downstream of these pathways by inhibiting the microtubular function required for cell migration and proliferation.20
If these differences in drugs and their targets were mechanistically relevant, one would expect that paclitaxel might be better suited than -olimus agents to preserve its antiproliferative effect in the diabetic state, independent of changes in upstream signalling. Because -olimus compounds act only on the PI-3-kinase pathway, their ability to inhibit SMC proliferation may be diminished under diabetic conditions due to downregulation of that pathway and increased traffic through the non-inhibited Ras/MAPK pathway.
Smooth-muscle Cells
Given the importance of inhibiting SMC proliferation and migration to the clinical benefit of DES, in vitro studies have sought to delineate the impact of paclitaxel and sirolimus on SMCs under diabetic conditions. Using an in vitro model of hyperglycaemia, Patterson et al.21 described the dysregulation of insulin-dependent signalling in rat aortic SMCs. Exposure of rat SMCs to hyperglycaemia resulted in significantly reduced Akt and insulin receptor substrate 1 (IRS-1) activation by insulin compared with that seen with normal glucose levels. In this model, sirolimus became 100-fold less potent in inhibiting SMC migration under hyperglycaemic conditions, with a dose-dependent decrease in inhibitory effect. By contrast, inhibition of SMC migration by paclitaxel remained virtually unchanged under high-glucose compared with normal-glucose conditions (see Figure 2).21 Another recent study has demonstrated that vascular endothelial cell growth factor (VEGF) protein levels in human coronary artery SMCs were dose-dependently increased by paclitaxel but decreased by sirolimus.22
Endothelial Cells
Although vascular SMC responses are central to restenosis, effective vessel healing post-stenting injury also requires increased endothelial cell migration and proliferation.23In vitro and in vivo evidence suggest that endothelial cell signalling is also altered by the diabetic state, and differentially influenced by -olimus drugs versus paclitaxel.
Using an in vivo model of pharmacologically controlled diabetes, Finn and colleagues24 have suggested that the differential effects of paclitaxel and -olimus compounds on endothelial cell signalling pathways in a diabetic state may be exacerbated by simultaneous exposure to the peroxisome proliferator-activated receptor gamma (PPARγ) agonist rosiglitazone, a drug commonly used to treat diabetes. In the presence of oral rosiglitazone, sirolimus-eluting stents (SES), but not paclitaxel-eluting stents (PES), showed a significant reduction in healing response, as demonstrated by reduced endothelial cell coverage 28 days after stenting (see Figure 3). Finn and colleagues suggest that this may be due to the inhibition by sirolimus of PPARγ-mediated VEGF transcription, which is regulated by mTOR. Therefore, blockade of mTOR by sirolimus may underlie the observed decrease in VEGF and endothelial cell strut coverage observed in animals receiving SES in conjunction with oral PPARγ agonists.24 While further study is needed, these results suggest the potential for mTOR inhibitors to further delay vascular healing in the presence of certain oral medications, due to convergence of molecular signalling.
Although zotarolimus, everolimus and other -olimus compounds also act via mTOR, corresponding data on their comparative effects under diabetic and non-diabetic states are not currently available. It is also important to note that PPARγ agonists are not used in all patients with diabetes and thus these in vitro and in vivo findings may not be generalised to all clinical settings.
Clinical Imaging Data
Volumetric ultrasound and angiographic studies demonstrate that intimal hyperplasia in PES-treated patients with diabetes is reduced relative to BMS-treated patients with diabetes to a level comparable to that seen in PES-treated patients without diabetes.25 As seen in the in vitro studies of hyperglycaemic SMCs,21 this supports the hypothesis that neointimal hyperplasia suppression is similar in patients with and without diabetes treated with PES.
By contrast, data following SES implantation suggest that angiographic late loss and restenosis may be higher in patients with versus patients without diabetes,26,27 although this difference is not statistically significant in all studies.28,29 These data contrast with a report of significantly lower eight-month neointimal volume in SES (0%) versus PES (7.5%) in patients with diabetes30 (compared with 30–35% typically seen after BMS treatment).25
However, optical coherence tomography (OCT) studies show that neointimal thickness is significantly greater in patients with diabetes compared with patients without diabetes treated with SES,31,32 while PES-treated patients had similar neointimal thickness regardless of diabetic status.32 These results are broadly consistent with the suggestion that SES may be less effective at inhibiting neointimal proliferation in patients with diabetes versus patients without diabetes.
Clinical Outcomes in Patients with Diabetes Treated with Drug-eluting stents
Clinical trial data support the prediction of in vitro studies by showing similar suppression of restenosis in PES-treated patients with diabetes versus patients without diabetes. In a pooled analysis of five randomised trials, the four-year target lesion revascularisation (TLR) rate was 24.7% in BMS-treated patients with diabetes compared with 12.4% in PES-treated patients with diabetes and 9.5% in PES-treated patients without diabetes (interaction p = not significant [NS]).15 Thus, PES reduced revascularisation compared with BMS in patients with diabetes to a rate similar to that seen in patients without diabetes. Moreover, while patients with diabetes versus patients without diabetes had significantly and expectedly higher four-year mortality rates than patients without diabetes, mortality was comparable in PES-treated (8.4%) and BMS-treated patients with diabetes (10.3%; p=0.61).15 Concordant results were observed in a newer generation of PES.33
These observations from PES randomised trials are further supported by results from the 7,492-patient TAXUS Peri-Approval Registry, A US Multicenter Safety Surveillance Program (ARRIVE) registry, which demonstrated that PES yield comparable TLR rates at one year (4.9 versus 5.3%; p=0.46) and two years (8.2 versus 7.7%; p=0.59) in patients with and patients without diabetes, respectively, even prior to adjustment for increased baseline risk in diabetes.34 Furthermore, data from the GlObal TAXUS Liberté RegistrY PrograM to SupPort WorldwIde CommerciAlization (OLYMPIA) registries show identical one-year unadjusted revascularisation rates of 3.1%.35
Figure 4A summarises 18 PES trials and registries including 10,535 patients with diabetes and 27,267 patients without diabetes. This model estimates a 12% relative increase (odds ratio [OR] 1.124, 95% confidence interval [CI] 1.015–1.244) in TLR for patients with medically treated diabetes compared with patients without diabetes.
By comparison, a parallel summary of -olimus stent trials and registries (see Figure 5A) shows a greater (46%) relative increase (OR 1.464, 95% CI 1.348–1.589) in revascularisation for patients with diabetes compared with patients without diabetes.
The impact of diabetes on restenosis tends to be greater in patients treated with insulin compared with oral hypoglycaemic agents alone. Therefore, we compared the available data on TLR rates for patients with diabetes treated with insulin or oral hypoglycaemic agents in the PES and SES data sets (see Figures 4B and 5B). The relative increase in TLR with insulin treatment was 31% in PES studies (OR 1.307, 95% CI 1.063–1.607) compared with 50% in -olimus studies (OR 1.501, 95% CI 1.247–1.806).
These analyses imply that while the presence of diabetes significantly increases revascularisation rates compared with patients without diabetes using either paclitaxel or -olimus stents, the magnitude of the increase was smaller in the paclitaxel studies. Multivariate modelling has also indicated that diabetes is an independent predictor of increased TLR among SES-treated patients,12,26,36,37 with no such indication in the PES experience.33,34 Similarly, insulin treatment was associated with a smaller relative increase in TLR compared with non-insulin-treated diabetes in the PES than in the SES studies.
Cross-comparison of Drug-eluting Stents
Unfortunately, there are only a few large head-to-head studies comparing different DES in patients with diabetes, with inconsistent findings between studies. These results are summarised in a plot of trials and registries that directly compared PES versus -olimuseluting stents in patients with diabetes (see Figure 6). It is important to note that most of the studies in this comparison reported the diabetes subgroup as a post hoc analysis; few randomised to PES versus -olimus exclusively in a population with diabetes with sufficient statistical power. Overall, this analysis shows a non-significant 7.3% lower revascularisation rate with PES versus -olimus (OR 0.927, 95% CI 0.823–1.044). Within the overall analysis, this finding is driven by a non-significant difference in favour of -olimus for the clinical trials (OR 1.250, 95% CI 0.977–1.599) but a significant difference in favour of PES for registries (OR 0.845, 95% CI 0.738–0.969).
Similar results are seen in a study-level meta-analysis of >12,000 patients with diabetes treated with PES or SES, reporting that among 13 randomised controlled trials and 16 registries overall, TLR and major adverse cardiac event (MACE) rates were similar between SES and PES, while among eight head-to-head registries, there were non-significant trends towards lower revascularisation (p=0.15) and MACE (p=0.06) in PES-treated patients with diabetes.38 On the other hand, a recent network meta-analysis including 3,852 patients with diabetes found no difference between PES and SES in cumulative TLR to four years post-stenting.14 These discrepancies emphasise the challenges of combining study-level data from sources with different patient composition, definitions of end-points and levels of end-point ascertainment/monitoring, particularly given the small and statistically underpowered diabetic subgroups in many of these studies.
Moreover, it is not clear whether any disparity seen between PES and SES is specific to SES or is potentially indicative of an -olimus class effect. In the SPIRIT III and IV studies comparing PES versus everolimus-eluting stent (EES), although TLR was significantly lower in the EES group in general populations, in patients with diabetes TLR was comparable between PES and EES.39,40 In the SPIRIT V diabetic randomised controlled trial, one-year TLR rates were numerically lower with the TAXUS Liberté PES (3.8%) versus XIENCE V/PROMUS (8.4%; p=0.16) (although late loss was significantly higher with TAXUS Liberté than XIENCE V/PROMUS).41 The COMPARE study demonstrated that while EES provided significantly lower MACE rates compared with the TAXUS Liberté stent overall and in most subgroups, this finding was not replicated in patients with diabetes, in which MACE outcomes were comparable between stent types.42 These results suggest potential benefits of implantation of EES rather than PES in many patients. Notably, however, PES appear to perform comparably to EES in the particularly challenging diabetic subgroup, despite the known increased risk in these patients.
A few studies have also compared PES versus zotarolimus-eluting stents (ZES) in patients with diabetes. In the Swedish Coronary Angiography and Angioplasty Registry (SCAAR) registry, the probability of restenosis in patients with diabetes versus patients without diabetes was significantly higher in patients treated with ZES or SES, but restenosis was similar regardless of diabetic status in patients treated with PES (see Figure 7), consistent with an -olimus class effect in diabetes.43
However, revascularisation rates were comparable for ZES-treated (6.9%) and PES-treated (5.8%; p=0.70) patients with diabetes in the ENDEAVOR IV study,44 and the TLR rate in ZES-treated patients with diabetes was significantly higher (13.7%) than that in SES-treated patients with diabetes (6.5%; p<0.05) in a 1,240-patient registry.45 Furthermore, the results from the SCAAR registry of over 8,200 patients with diabetes demonstrated that the probability of restenosis two years post-stenting was significantly higher (twice as high) with ZES compared with either PES or SES, with no significant differences between SES and PES.43
Thus, although studies of single stent types generally show less increase in revascularisation rates for PES than -olimus stents (for patients with versus patients without diabetes, and for insulin versus hypoglycaemic oral agents), in keeping with the in vitro findings, the available head-to-head studies show mixed results that neither confirm nor deny a significant attenuation in the diabetic revascularisation ‘penalty’ for PES compared with -olimus DES. The question of differential endothelial cell migration and proliferation between the two drugs is more difficult to study clinically, and may require more detailed studies of strut coverage (e.g. OCT) or endothelial function (vascular reactivity) for further insight. It is also important to remember that in addition to mechanisms of drug action, stent-platform design (e.g. strut thickness, polymer and homogeneity of drug distribution) may have an effect on the relative efficacy of different DES in patients with diabetes and patients without diabetes.
Summary – Treating Coronary Artery Stenosis in the Diabetic Patient
Patients with diabetes present one of the most challenging subgroups for percutaneous coronary intervention (PCI). DES provide substantial benefits over both BMS and simple balloon angioplasty in patients with diabetes, without increasing mortality. However, the long-term mortality of patients with diabetes is increased compared with patients without diabetes regardless of treatment modality and continues to call for medical therapy to prevent the increased pace of diabetic vascular disease and improve outcomes following PCI.1,46 This bench-to-bedside review has explored whether there are differences between -olimus and paclitaxel-based DES in patients with diabetes.
The in vitro results suggest that paclitaxel, by virtue of its more distal microtubular mode of action, may be better able to maintain its inhibition of SMC migration and proliferation in the diabetic condition, while -olimus compounds may partially lose efficacy in the diabetic compared with the non-diabetic state. Additionally, paclitaxel drugs do not appear to interfere with endothelial healing in a diabetic model, while sirolimus seems to produce incremental endothelial inhibition (which is further increased by concomitant PPARγ agonist administration) due to signalling convergence of mTOR inhibition in endothelial repair pathways. The impairment in endothelialisation with SES (but not PES or BMS) in the presence of a glitazone is suggestive, but requires clinical confirmation.
These bench findings are mirrored in studies of single stent types that show little if any increase in neointimal growth and less impact on clinical TLR between patients with diabetes and patients without diabetes for PES compared with larger increases in TLR for -olimus DES. The same pattern is seen comparing insulin-treated versus non-insulin- treated patients with diabetes, with smaller increases in TLR for PES than for -olimus DES. However, clinical studies directly comparing paclitaxel and -olimus stents in diabetic subgroups show mixed outcomes, and caution must be used in interpreting the results of studies with different inclusion criteria, study designs and diabetes definition. Operator stent choice in the registry studies (based on the expected efficacy of each stent type in patients with diabetes) may also have an impact on the apparent outcomes. Indeed, the results of the SYNergy between percutaneous coronary intervention with TAXus and cardiac surgery (SYNTAX) study have demonstrated that assessment of baseline patient risk is important in order to better assess the potential efficacy of different revascularisation procedures in patients with diabetes (i.e. stenting versus surgery).47 Additional evidence from adequately powered randomised trials will thus be required to confirm whether the in vitro, animal model and individual DES study data comparing PES with -olimus stents will be translated into a compelling net clinical benefit in the challenging subgroup of patients with diabetes. Ôûá