Percutaneous transluminal coronary angioplasty (PTCA) is an established myocardial revascularization procedure; however, restenosis rate ranges from 30% to 60% (1), and in unstable patients the risk of acute complications ranges from 5% to 30% (2- 3).

Preprocedural identification of low-risk and high-risk patients who might benefit from additional procedures (4- 5) would be desirable. However, the predictors of acute complications (3,6- 12) and restenosis (13- 18) considered so far have a low predictive value (7,9,18), are available only after the procedure (11,17) or are not easily applicable in clinical practice (10- 12,15- 16).

In vitro enhanced cytokine synthesis by peripheral blood monocytes before PTCA has been found to predict late lumen loss, suggesting that preprocedural activation of inflammatory cells may play a role in the modulation of vessel wall response to the injury induced by balloon PTCA (16). This possibility is supported by experimental and clinical studies showing that acute phase reactants and proinflammatory cytokines promote leukocyte, endothelial and smooth muscle cell activation, resulting in an increase in procoagulant activity (19), metalloproteinase release (20) and neointimal proliferation (21).

C-reactive protein (CRP), serum amyloid A protein (SAA) and fibrinogen are easily measurable acute-phase reactants, which are synthesized in response to proinflammatory cytokines (22). They have been consistently associated with prognosis in ischemic heart disease (23- 26). Therefore, we investigated the short- and long-term prognostic value of preprocedural serum levels of CRP, SAA and fibrinogen in patients with stable and unstable angina undergoing single-vessel PTCA.

The study population consisted of 121 of 219 consecutive patients who underwent PTCA on a single nonocclusive coronary stenosis; 52 had stable angina and 69 had unstable angina (Table le1).

Exclusion criteria were: myocardial infarction (MI) within three months (28 patients), multilesion PTCA (15 patients), total occlusion (8 patients), previous PTCA or bypass surgery (27 patients), left ventricular ejection fraction <30% (12 patients), left bundle branch block (3 patients) and intercurrent inflammatory conditions known to be associated with an acute phase response (5 patients). At the time of PTCA, all patients were on oral aspirin.

The protocol was approved by the Ethics Committee of the Catholic University of Rome; all patients gave informed consent.

Peripheral blood samples for measurement of CRP, SAA and fibrinogen were taken immediately before PTCA. Coded plasma samples were stored at −70°C and analyzed in a single batch at the end of the study; thus, patient management was independent of these results.

C-reactive protein and SAA were assayed by an automated monoclonal antibody solid phase sandwich-type enzyme immunoassay on the Abbott IM_X instrument (Abbott Laboratories, North Chicago, Illinois) (27- 28). Fibrinogen was measured by the thrombin time method, as described by Clauss (29).

For data analysis, the overall population was grouped in tertiles according to preprocedural levels of CRP, SAA and fibrinogen. Moreover, elevated preprocedural levels were defined as: those above 0.3 mg/dl for CRP levels (i.e., 90th percentile of the normal distribution) (23), above 0.5 mg/dl for SAA levels (i.e., 82th percentile of the normal distribution) (23) and above 350 mg/dl for fibrinogen (26).

The procedure was performed using a steerable balloon catheter system via the femoral route. All patients received heparin at the dose required to maintain the activated clotting time above 300 s throughout the procedure.

Coronary angiograms were independently reviewed by two expert angiographers who were unaware of the patients’ clinical and analytic data. Angiographic lesion morphology before PTCA was categorized according to Ambrose et al. (30) and to a modified scheme of American College of Cardiology/American Heart Association Task Force classification (9). Minimal luminal diameter, reference diameter, percent diameter stenosis, acute gain and balloon/vessel ratio were assessed by computerized quantitative angiography (Medis, Nuenen, Netherlands). Procedural angiograms were also analyzed for Thrombolysis in Myocardial Infarction (TIMI) flow grade (31), the appearance of intracoronary thrombi (intraluminal filling defects or contrast medium staining within the lumen) and luminal dissection (9). Percutaneous transluminal coronary angioplasty was considered successful if the final percent diameter stenosis was less than 50% with TIMI 3 flow in the absence of death, recurrent ischemia, MI (creatine kinase increase to more than two times the upper limit of normal with or without evidence of new Q-waves), need of bailout stenting or urgent coronary bypass graft surgery (CABG) during the hospital period. Angiographic restenosis was defined as more than 50% stenosis in a previously successfully dilated lesion.

After the procedure, all patients had creatine kinase measurements every 6 h for 24 h and daily recording of symptoms and electrocardiogram (ECG) until hospital discharge (4 ± 2 days after the procedure). All patients were discharged on diltiazem and aspirin; other drugs were used if clinically indicated. Clinical examinations and treadmill stress tests were performed at 1, 3, 6 and 12 months. Coronary angiography was repeated in patients with evidence of clinical restenosis as defined below.

The following intraprocedural and predischarge complications were considered as early adverse events:

At follow-up, the primary end point was the clinical evidence of restenosis within one year after hospital discharge. Clinical restenosis was defined as the recurrence or worsening of ischemic symptoms (typical angina, MI or death) or ischemia at exercise testing (≥1 mm ST segment depression). Only patients with clinical evidence of restenosis underwent repeat coronary angiography.

The need for repeat coronary revascularization, PTCA or CABG, was also evaluated.

Continuous variables with normal distribution are expressed as mean ± SD and compared by t tests; CRP, SAA and fibrinogen values, not normally distributed, are presented as median and range and compared by Mann-Whitney U test. Correlations were determined using Spearman’s rank correlation test. Categorical variables were compared by Fisher exact test or chi-square statistics, as indicated; when rates of events were calculated for each tertile of distribution of a continuous variable (CRP, SAA and fibrinogen levels), the test for linear association (Mantel-Haenszel) was also applied. Logistic regression analysis was performed to determine predictors of early and late adverse events (univariate and forward stepwise analysis). The chi-square model was used to test the significance of the coefficients for all the terms in the model. To verify whether the model fit the data reasonably well, the Hosmer-Lemeshow Goodness-of-Fit statistics were computed. Event free survival at follow-up was analyzed by Cox proportional hazards regression model. Tertile distribution of CRP, SAA and fibrinogen levels, as well as several clinical and angiographic variables listed in (Table le1), were included in the regression analysis. In addition, CRP, SAA and fibrinogen levels were included in the regression analysis as continuous variables after their logarithmic transformation in order to obtain a nearly normal distribution.

The statistical analysis was performed using the Statistical Package for Social Sciences software (SPSS 8.0 for Windows, SPSS Inc., Chicago, Illinois). A p value < 0.05 was considered statistically significant.

Clinical characteristics, angiographic findings and procedural variables are summarized in (Table le1). Before PTCA, elevated levels of CRP (>0.3 mg/dl), SAA (>0.5 mg/dl) and fibrinogen (>350 mg/dl) were observed in 29%, 21% and 14%, respectively, of patients with stable angina and in 78%, 68% and 49%, respectively, of patients with unstable angina (all p < 0.001 vs. stable angina) (Table le2). Serum amyloid A protein levels were closely correlated with CRP levels (r = 0.91, p < 0.0001), and fibrinogen levels were weakly correlated with CRP (r = 0.43, p < 0.001) and SAA levels (r = 0.48, p < 0.001).

Fifteen early adverse events were observed in the whole population (12%), 2 in stable and 13 in unstable patients (p = 0.01). Twelve acute or threatened acute occlusions occurred during the procedure. The remaining three adverse events occurred before hospital discharge: two patients developed an acute MI and one patient had recurrence of rest angina. Percutaneous transluminal coronary angioplasty failed in 10 patients: urgent CABG in 5 patients, bailout stenting in 2 patients, in-hospital recurrence of ischemia in 3 patients. Early adverse events, treatments and outcome are reported in detail in (Table le3).

Among the 52 patients, two (4%) had an early adverse event: one had an acute occlusion, the other had a threatened acute occlusion. They had elevated levels of CRP (2.2 and 18.7 mg/dl) and SAA (1.7 and 20.5 mg/dl) and both were in the top tertile of CRP and SAA levels (Figure 1).

Among the 69 patients, 13 (19%) had an early adverse event. An acute occlusion occurred in 7, threatened acute occlusions during the procedure in 3 and early recurrence of ischemia before hospital discharge in 3. All unstable angina patients with early adverse events had elevated levels of CRP (median 2.1, range 0.47 to 11.9 mg/dl) and SAA (median 1.17, range 0.51 to 15.9 mg/dl). The incidence of early adverse events increased from 0% in the bottom tertile to 30% in the top tertile of CRP levels (p = 0.014); a similar trend was observed for SAA levels (from 0% to 27%, p = 0.045) (Figure 1). Fibrinogen levels did not discriminate against patients with or without early adverse events (Figure 1).

One year follow up was completed in all 52 patients with stable angina and in 59 patients with unstable angina (10 patients with failed PTCA were excluded from follow up).

Clinical restenosis occurred in 51 patients (46%), as reported in detail in (Table le4). Repeat angiography confirmed a severe (>75%) target lesion restenosis in all but one patient who did not exhibit any critical stenosis.

Clinical restenosis rate was similar in patients with stable or unstable angina (21/52, 40% vs. 30/59, 51%, p = 0.27). Clinical restenosis was significantly higher among patients with elevated levels of CRP (37/59, 63% vs. 14/52, 27%, p < 0.001), SAA (34/57, 60% vs. 17/54, 32%, p = 0.003) and fibrinogen (23/36, 64% vs. 28/75, 37%, p = 0.009).

Restenosis rates were 30% and 33% in stable patients with normal levels of CRP and fibrinogen and 67% and 86%, respectively, in those with elevated levels (p = 0.014 and p = 0.013). Incidence of restenosis increased from 33% in the I tertile to 86% in the III tertile for both CRP levels and fibrinogen (p = 0.046) (Figure 2). A similar trend was observed for SAA levels (p = 0.13) (Figure 2).

Restenosis rates were 20% and 32% in unstable patients with normal levels of CRP and SAA and 61% and 62%, respectively, in those with elevated levels (p = 0.006 and p = 0.024). An increasing incidence of restenosis was observed from 25% in the I tertile to 72% in the III tertile of CRP (p = 0.004); a similar trend was observed for SAA (from 23% to 67%, p = 0.010) (Figure 2). Fibrinogen levels were not associated with the incidence of restenosis (Figure 2).

Tertiles of CRP and SAA levels, unstable angina, hypertension and female gender were associated with a higher risk of early adverse events at the univariate regression analysis. However, CRP tertiles (RR = 12.2, confidence interval [CI] = 3.0 to 50.2, III tertile vs. the remaining two p < 0.001), female gender (RR 4.1, CI = 1.1 to 14.7, p = 0.033) and hypertension (RR = 4.3, CI = 1.0 to 18.5, p = 0.046) were the only independent predictors of early adverse events (chi-square model = 26.7, df [degrees of freedom] = 3, p < 0.001) (Table le5). The risk predicted by our logistic regression model was well correlated with observed early adverse events (87% of correct classification; Hosmer-Lemeshow Goodness-of-Fit, p = 0.939).

When acute-phase proteins were included in the regression analysis as continuous variables, CRP levels (RR for log [CRP] = 21.9, CI = 4.7 to 102.1, p < 0.001) were the only independent predictors of early adverse events (chi-square model = 27.4, df = 1, p < 0.001); after exclusion of other inflammatory markers, SAA (RR for log [SAA] = 8.1, CI = 2.5 to 26.6, p < 0.001), but not fibrinogen levels, independently predicted early adverse events. C-reactive protein levels (RR = 10.4, CI = 1.4 to 79.6, III tertile vs. I tertile, p = 0.02) were also an independent predictor of the in-hospital occurrence of death, MI, need of urgent CABG or repeat revascularization.

Tertiles of CRP, SAA and fibrinogen levels before PTCA, multivessel disease and residual diameter stenosis were associated with increased risk of clinical restenosis at the univariate analysis. However, CRP (RR = 6.2, CI = 2.0 to 18.7, III tertile vs. I tertile, p = 0.001) and residual stenosis (RR = 3.2, CI = 1.3 to 7.5, >30% vs. ≤30% stenosis, p = 0.007) were the only independent predictors of restenosis (chi-square model = 23.6, df = 3, p < 0.001) (Table le6). The risk predicted by our multivariate logistic analysis was closely correlated with observed events (percentage of correct classification = 70%; Hosmer-Lemeshow Goodness-of-Fit, p = 0.968). Of note is that elevated CRP levels, compared with low levels, increased the restenosis rate from 18% to 44% in patients with ≤30% residual stenosis and from 37% to 77% in patients with >30% residual stenosis (Figure 3). When acute-phase proteins and residual stenosis were analyzed as continuous variables, SAA levels (RR for log [SAA] = 6.0, CI = 2.3 to 16.2, p < 0.001), acute gain (RR = 0.3, CI = 0.1 to 0.8, p = 0.011) and multivessel disease (RR = 2.7, CI = 1.0 to 7.0, p = 0.042) were the only independent predictors of clinical restenosis (chi-square model = 25.6, df = 3, p < 0.001); after exclusion of other inflammatory markers, CRP [RR for log [CRP] = 2.9, CI = 1.4 to 6.3, p = 0.004], but not fibrinogen levels, were also independent predictors of restenosis. Preprocedural CRP levels (RR = 4.9, CI = 1.6 to 14.4, III tertile vs. I tertile, p = 0.008) were also the most powerful predictor of major adverse cardiac events (e.g., death, MI or need for repeat revascularization) at follow-up.

Cox proportional-hazard model was used in order to evaluate timing of adverse events at follow-up; CRP and residual stenosis were the most powerful independent predictors of restenosis and major cardiac adverse events at follow-up (Figure 4).

To our knowledge, this prospective study is the first to provide evidence that, in a consecutive group of patients undergoing single vessel PTCA, early complications and late restenosis can be predicted with a reasonable accuracy by measurements easily obtained before the procedure.

Normal CRP levels had a 100% negative predictive value for early adverse events and identified a subset of patients (43% of the whole population) who did not require additional treatments until hospital discharge. Inconsistent and weak correlations with early complications following PTCA were reported for female gender (8), extreme age (7), diabetes (8), multivessel disease (6,8), lesion characteristics (8- 10) and hemostatic variables (11- 12). Evidence of intracoronary thrombus was found to be a more consistent predictor of early adverse events (6,8,10), but its practical value is limited by its low prevalence in the baseline angiogram (10). In clinical practice, unstable angina is the most important predictor of acute complications following PTCA (2- 3,6); however, our observation that serum CRP levels are even stronger predictors of early adverse events suggests that the degree of activation of inflammatory cells is a more important determinant of early outcome after PTCA than clinical instability. Elevated SAA levels had a similar prognostic value to CRP levels. Conversely, fibrinogen failed to show a prognostic value in unstable angina patients and for early adverse events. This may be due to a smaller dynamic range of fibrinogen (lower increase and longer half-life than CRP and SAA) and to the fact that fibrinogen levels depend not only on the production rate but also on its consumption rate, influenced by anticoagulant drugs.

Normal CRP levels had a negative predictive value of 73% and identified a subset of patients (47% of our population) showing, in the presence of ≤30% residual stenosis, a restenosis rate as low as 18%. Conversely, high CRP levels had a positive predictive value of 63% and identified a subset of patients (53% of our population) with an unacceptable restenosis rate, regardless of the intraprocedural angiographic result (restenosis ranging from 44% to 77% according to residual stenosis).

Unstable angina (5,13), diabetes (13), serum levels of fibrinogen (17), lipoprotein (a) (18), indexes of fibrinolysis (15) and of platelet function (15) and a number of lesion- and procedural-related variables (13- 14) were reported to predict restenosis, but with a low predictive value. Our results confirm the prognostic value of residual stenosis, but they also show that CRP levels before the procedure are the most powerful predictor of restenosis, even in patients with a good angiographic result.

Acute complications following PTCA are commonly due to thrombus formation or vessel dissection (2). Activation of inflammatory cells may promote acute complications by increasing procoagulant activity (19) and by increasing the risk of dissection and of plaque hemorrhage, through an enhanced synthesis of matrix metalloproteinases (20). Accordingly, enhanced intraprocedural thrombin generation (11) and preprocedural platelet activation (12) have been reported in patients undergoing acute occlusion and acute ischemic events after PTCA.

Restenosis has often been regarded as an inevitable local healing process following PTCA. However, the bimodal distribution of late lumen loss, demonstrated following balloon PTCA (32) and stenting (33), as well as the clustering of restenosis in patients with multivessel PTCA (34), suggests that restenosis is not necessarily a uniform vessel wall response to balloon injury but may occur in some lesions and in some patients but not in others. The independent predictive value of residual stenosis and acute-phase reactants suggests the presence of at least two different components operating in the restenotic process: a mechanical component mainly operating in patients with a suboptimal dilation of the lesion at the end of PTCA and an inflammatory component, mainly operating in patients with preprocedural elevated levels of acute phase reactants and probably causing an enhanced vessel response to balloon injury.

Our results are consistent with the recent observation that enhanced cytokine synthesis by peripheral blood monocytes evaluated in vitro before PTCA predicts late lumen loss (16) and suggest that preprocedural activation of inflammatory cells may influence intimal hyperplasia and vascular remodeling.

The excellent negative predictive value of preprocedural levels of CRP or SAA found in this study needs to be considered cautiously due to the relatively small number of patients (and events). However, the prognostic value of these markers, if confirmed in larger clinical studies, could contribute to optimizing therapeutic resources in the complex scenario of interventional cardiology (4- 5).

Another limitation of our study is the lack of repeat angiography in asymptomatic patients. However, an increasing number of studies are focusing on clinical more than on angiographic restenosis. Accordingly, we were more interested in risk stratification of patients undergoing balloon PTCA rather than the pathophysiologic mechanisms of restenosis. Finally, the role of inflammation in restenosis might be assessed more easily after stenting, which is not affected by the mechanical component of restenosis (e.g., suboptimal dilation and vessel recoil). In a recent study, CRP levels were found to predict restenosis following stent implantation (35).

Our study shows that serum levels of CRP before PTCA provide a more powerful predictor of both acute complications and clinical restenosis than clinical presentation and other risk factors considered thus far. Low CRP levels before the procedure may help identify patients with a good outcome when treated by conventional balloon PTCA. Therefore, for a better resource administration in interventional cardiology and until specific causes of acute complications and restenosis are identified, complex and expensive therapeutic tools, such as anti-glycoprotein IIb/IIIa agents (4) and stenting (5), may be reserved for patients with high levels of CRP who are at the highest risk of in-hospital adverse events and late restenosis.