Coronary artery stenting [10, 11] provides a rigid endovascular lattice that prevents both elastic recoil and vasospasm at the treated site. Restenosis within a stent results almost exclusively from intimal hyperplasia; thus, a study in patients with stents provides a potent model for evaluating the role of smooth muscle cell hyperplasia in restenosis independent of any contribution from elastic recoil or vasospasm. The effects of metabolic disorders, such as diabetes, on intimal hyperplasia and the contribution of intimal hyperplasia to restenosis can be examined directly.

Further complicating the study of the relation between metabolic disorders and restenosis has been the reliance on traditional concepts of restenosis as a dichotomous event [12, 13]. However, using a continuous geometric model in which restenosis is separated into its constituent parts (acute gain and late loss of minimum lumen diameter [14]), one can examine the effect of disease states such as diabetes on specific components of restenosis.

Methods

Top Methods Results Discussion Author & Article Info References

Patient Sample

Between June 1988 and July 1991, 250 Palmaz-Schatz stents (Johnson & Johnson, Interventional Systems, Warren, New Jersey) were placed in 220 patients as part of a multicenter study of this investigational device. Patients were selected for stenting based on morphologic features of the lesion that suggested an increased risk for a suboptimal result or restenosis after conventional balloon angioplasty. The protocol was approved by the Committee on Clinical Investigation, Beth Israel Hospital, Boston, Massachusetts, and all patients gave informed consent. The inclusion and exclusion criteria have been described previously [15]. Briefly, all patients had objective evidence of myocardial ischemia, as well as stenosis of 70% or more in either native vessels or saphenous vein grafts, with a lesion length of less than 15 mm and a reference artery diameter of 3 mm or more.

Medical Therapy

All patients were pretreated with aspirin, 325 mg/d; dipyridamole, 200 mg/d; and dextran, 200 mL. Intravenous heparin, 10 000 units, was administered before balloon dilatation with additional heparin given to maintain an activated clotting time of more than 300 seconds. Therapeutic heparinization (activated partial thromboplastin time 2 to 2.5 times the control value) was continued until warfarin therapy prolonged the prothrombin time to 16 to 18 seconds; warfarin therapy was continued for 8 weeks to prevent subacute stent thrombosis. Aspirin and dipyridamole were continued indefinitely.

Coronary Stents

The Palmaz-Schatz stent is a 15-mm, articulated, slotted, stainless steel tube mounted on a conventional angioplasty balloon. By November 1991, more than 1500 Palmaz-Schatz stents had been placed in coronary arteries and saphenous vein grafts as part of this multicenter study. The procedure for stent deployment [11, 15] entails predilation with a conventional angioplasty balloon, following which the stent (mounted on an angioplasty balloon) is positioned across the lesion. The balloon is inflated to expand the stent, permanently deploying it against the vessel wall. The stent was dilated further with a larger balloon to match the stent diameter to that of the adjacent reference artery.

Angiographic Follow-up

Follow-up angiograms were obtained in 189 (82%) of 230 "eligible" lesions. Lesions became eligible for follow-up study 4 months after successful stent placement.

Angiographic Analysis and the Geometric Model

Baseline angiograms were obtained in multiple projections before intervention. Identical angiographic views were repeated immediately after stent placement and at the follow-up study. Each lesion and its proximal and distal reference segments were measured before and after stent placement, as well as at follow-up, using digital calipers (Fowler Ultra-Cal II, Sylvac, Zurich, Switzerland) from an optically magnified image. Absolute lumen diameters were obtained by comparing measured values with known absolute values from the calipered guiding catheter. Percent stenosis was calculated by subtracting the ratio of the minimum lumen diameter (MLD) to the reference artery diameter from unity (1 –MLD/reference artery). Minimum lumen diameter (mm), reference artery diameter (mm), and percent stenosis were obtained for all lesions before and after stenting and at follow-up.

Angiographic restenosis was analyzed per lesion according to a continuous geometric model first described by Kuntz [14]:

Acute gain = MLD (post-procedure) –MLD (pre-procedure)

Late loss = MLD (post-procedure) –MLD (at follow-up)

Loss index = late loss ÷ acute gain or

MLD (post-procedure) –MLD (at follow-up)

MLD (post-procedure) –MLD (pre-procedure)

Statistical Analysis

Data are expressed as mean ±SD. Categorical variables were compared using the chi-square test, and continuous variables were compared using the Student t-test. A P value of 0.05 or less was considered statistically significant.

Results

Top Methods Results Discussion Author & Article Info References

Patient Characteristics

Two hundred fifty stents were successfully placed in 216 of 220 patients, 37 (17%) of whom had diabetes mellitus. Thirty-five of 37 patients (94%) were classified as having type II diabetes. At the time of stent placement, 15 of 37 patients (41%) were receiving oral hypoglycemic agents, 11 of 37 (30%) were receiving insulin, and 11 of 37 (30%) were receiving dietary management alone. Follow-up angiograms were obtained for 189 of 230 stents (82%): 29 of 40 stents (73%) in diabetic patients and 160 of 190 stents (84%) in nondiabetic patients. Based on our analysis of lesions eligible for follow-up, diabetic and nondiabetic patients did not differ significantly with regard to age, sex, history of cigarette smoking, hypertension, hypercholesterolemia, family history of coronary artery disease, or previous coronary intervention (Table 1). The incidence of clinical events that might preclude repeated angiographic study did not differ between diabetic patients (two deaths and one bypass surgery) and nondiabetic patients (three deaths, one subacute thrombosis, and one bypass surgery). Because the end point of our analysis was angiographic restenosis, a lesion-based analysis was done. In studies of clinical restenosis, patient-based analysis is more appropriate.

Baseline Angiography

The distribution of vessels in which stents were placed was similar for diabetic and nondiabetic patients (Table 2). Vessels in patients with diabetes mellitus did not differ from those in nondiabetic patients regarding size (reference artery diameter, 3.31 ± 0.65 mm compared with 3.39 ± 0.65; P > 0.2) Table 3 or percent stenosis before treatment. After stent placement, the mean residual percent stenosis and minimum lumen diameter were similar in both groups. Thus, the acute gain did not differ significantly between stents placed in diabetic and nondiabetic patients (Figure 1).

View larger version (30K): [in this window] [in a new window]   Figure 1. Restenosis in diabetic and nondiabetic patients according to the terms of the geometric model. The model is described in the Methods section. The top of the bar shows the mean reference artery diameter, and the black area represents the minimum lumen diameter before treatment (pre-procedure). The incremental gain in lumen diameter before and immediately after stenting represents the acute gain. Late loss, calculated as the difference in minimum lumen diameter immediately after placement and at follow-up (hatched area), was significantly greater in diabetic patients.

Follow-up Angiography

The time to angiographic follow-up did not differ for stents placed in diabetic and nondiabetic patients (176 ± 49 days compared with 175 ± 49 days; P > 0.2). At follow-up, however, the minimum lumen diameter of stents placed in diabetic patients was significantly smaller than that of stents placed in nondiabetic patients (1.66 ± 1.18 mm compared with 2.24 ± 0.93 mm; P = 0.004), with greater corresponding percent stenosis (see Table 3). Therefore, late loss in stents placed in diabetic patients was significantly greater than in those placed in nondiabetic patients (1.66 ± 1.28 mm compared with 1.23 ± 0.97 mm; P = 0.04) (see Table 3), explaining the smaller late lumen diameter seen in stents placed in diabetic patients. Because the amount of late loss is known to depend on the amount of acute gain, late loss can be normalized for acute gain. The resulting loss index was significantly greater in stents placed in diabetic patients (Figure 2). This increase in the loss index seen in stents in diabetic patients accounts for the significantly higher incidence of restenosis (55% compared with 20%; P = 0.001) among stents placed in diabetic patients when restenosis is expressed as a traditional dichotomous outcome: stenosis of 50% or more at follow-up (Figure 3). Although a greater number of stents was placed in saphenous vein grafts in diabetic patients, the restenosis rate did not differ statistically between native vessels and vein grafts (27% compared with 21%; P = 0.30).

View larger version (19K): [in this window] [in a new window]   Figure 2. Loss index in diabetic and nondiabetic patients. The loss index is calculated as late loss divided by acute gain.

View larger version (17K): [in this window] [in a new window]   Figure 3. Cumulative distribution of percent stenosis in stents at follow-up for diabetic and nondiabetic patients. Angiographic restenosis (50% diameter stenosis) was significantly greater in diabetic patients.

Discussion

Top Methods Results Discussion Author & Article Info References

Diabetes mellitus often results in angiopathy of both large and small vessels [1, 16, 17]. Patients with diabetes also exhibit abnormal wound healing [18, 19], which involves three phases—inflammation, granulation, and extracellular matrix formation [20, 21]—that bear a superficial similarity to the healing process in a vessel after coronary angioplasty.

The anatomic derangements that follow balloon dilatation have been further characterized [9] by postmortem [7, 22], angioscopic [23], and experimental studies. Endothelial denudation, platelet deposition, mural thrombosis, and medial fracture are commonly associated with balloon injury. Accumulating data [7, 24, 25] suggest that this proliferative response becomes quiescent several months after intervention.

Current evidence strongly supports the concept that intimal smooth muscle cell hyperplasia is the predominant histologic finding in restenotic lesions [26, 27]. Based on traditional binary definitions, restenosis is considered an "all or none" phenomenon, but analysis of intimal hyperplasia shows that it is ubiquitous after arterial injury and occurs even in patients who do not manifest clinical evidence of restenosis [14, 28]. To quantify this hyperplastic response and to determine how pathologic states such as diabetes affect its magnitude, we developed a novel, continuous model [14]. Use of this model rather than the traditional dichotomous definition of restenosis [13] allows a more quantitative analysis.

According to this model, restenosis can be separated into its constituent components of acute gain and late loss, where acute gain refers to the pre- to postintervention increase in minimum vessel diameter. In our study, no significant differences in acute gain were found between the diabetic and nondiabetic groups. Small and diffusely diseased vessels, as well as a suboptimal postprocedure outcome, would be predictors of increased restenosis rates [15, 29, 30], but reference artery diameter, lesion length, and acute gain were not different for lesions in diabetic patients when compared with those in nondiabetic patients. Although a greater percentage of lesions in the diabetic cohort occurred in saphenous vein grafts (54% compared with 33%; P = 0.07), we have previously shown that, unlike in conventional balloon angioplasty, the restenosis rate for stented saphenous vein grafts and native coronary arteries are almost identical [15]. In our analysis, native vessels and vein grafts did not differ significantly in restenosis rate. Therefore, the greater number of vein grafts in the diabetic cohort should not confound the analysis of restenosis in these two groups. Our study suggests that the higher rate of restenosis in diabetic patients cannot be attributed to inherent differences in vessel size, lesion length or severity, or immediate improvement after the procedure.

Late loss (defined as the reduction in minimum lumen diameter from immediately after the procedure to 6-month follow-up) was significantly greater in stents placed in diabetic patients. In patients treated with conventional balloon angioplasty, this decrement in lumen diameter may be due in part to vessel recoil [8] or spasm. By providing a rigid, vascular scaffolding, however, the stent prevents both elastic vascular recoil and spasm. Thus, an analysis of the loss index in stents provides a putative quantitative measure of intimal hyperplasia, suggesting that the markedly increased restenosis rate seen in diabetic patients after stenting may be related to greater smooth muscle proliferation within the stent. This finding is consistent with a histologic analysis of resected atheromata from patients with peripheral vascular disease [31], which showed an increased number of smooth muscle cells in patients with diabetes.

There are clear pathophysiologic precedents for why this intimal response might be greater in diabetic patients. After balloon injury, vascular smooth muscle cells undergo profound phenotypic changes [20] that are driven by an increase in responsiveness to several growth factors, including platelet-derived growth factor and insulin-like growth factor I [32-34]. Insulin itself may stimulate proliferation of arterial smooth muscle cells [16], and sera from patients with poorly controlled non-insulin-dependent diabetes have been shown to stimulate smooth muscle proliferation to a greater degree than sera from patients with well-controlled diabetes or from normoglycemic persons [35].

Our study suggests that the greater clinical restenosis rate observed in diabetic patients treated with coronary stents is due to enhanced late loss of luminal diameter, which may be caused by more pronounced smooth muscle hyperplasia within the stent. However, our study has several important limitations. First, the stent model of restenosis does not assess whether diabetes also affects other dynamic components of restenosis, such as vessel recoil or spasm. Thus, these findings cannot necessarily be extrapolated to restenosis occurring after conventional balloon angioplasty. Second, although accumulating data indicate that intimal hyperplasia is the pathologic mechanism underlying increased late loss after stenting, we did not obtain histologic confirmation in our patients. Therefore, our data only suggest that the greater late loss is due to hyperplasia in diabetic patients. In addition, the lack of histologic specimens precludes assessment of other contributors to intimal hyperplasia, such as ground matrix formation. Finally, we did not address the question of which mediator or mediators of intimal hyperplasia cause the exuberant proliferation seen in diabetic patients, nor did we assess the effects of insulin levels, diabetes type, or diabetic control on this process.