Wan–Jun CHENG, Shi–Wei YANG, Fei GAO, Yong–He GUO, Zhi–Jian WANG, Yu–Jie ZHOU

Beijing Anzhen Hospital Affiliated to Capital Medical University, Beijing Institute of Heart, Lung and Blood Vessel Disease, The Key Laboratory of Remodeling-related Cardiovascular Disease, Ministry of Education, Beijing, China

Abstract Objective The aim is to evaluate the association between baseline platelet count (PC) and severe adverse outcomes following percutaneous coronary intervention (PCI) in current real-world practice. Methods A total of 18,788 patients underwent PCI with drug-eluting stents constituted the study population. Patients were categorized as having low (＜ 150 × 1000/μL), normal (150–300 × 1000/μL), and high(≥ 300 × 1000/μL) baseline PC. The primary endpoints included in-hospital and follow-up all-cause mortality. The secondary endpoint was major bleeding requiring a blood transfusion. Results In-hospital mortality rates for patients with low, normal, and high baseline PC were 0.6%, 0.4%, and 0.4%, respectively (P = 0.259). Similarly, mortality rates during long-term follow-up (median 23.8 months) for patients with low, normal, and high baseline PC were 0.9%, 0.6%, and 0.7%, respectively (P = 0.079). After multivariate adjustment, patients with low or high baseline PC tended to have similar risks for both in-hospital and follow-up mortality compared with the normal group. Subgroup analyses failed to demonstrate an independent prognostic value of baseline PC in specific population groups except patients who undwent transfemoral PCI. There was also no significant difference in the incidence of major bleeding requiring a blood transfusion in the low, normal,and high groups (0.5%, 0.3%, and 0.3%, respectively; P = 0.320). After multivariate adjustment, low or high baseline PC did not significantly increase the risk of major bleeding. Conclusion There is no significant association between baseline PC and severe adverse outcomes following PCI in current real-world practice.

Keywords: Major bleeding; Mortality; Percutaneous coronary intervention; Platelet count

1 Introduction

Thrombosis and hemorrhage represent the main challenges of myocardial revascularization.[1,2]Platelets play a key role in the pathophysiological process of both thrombosis and hemorrhage.[3,4]An abnormal increase (thrombocytosis) or decrease (thrombocytopenia) in platelets may cause defective formation of hemostatic plugs and bleeding.[5,6]Accordingly, such patients were excluded from the vast majority of clinical trials, given the potentially increased risks. Few pooled post-hoc analyses[7–15]and cohort studies[13–15]drew inconsistent conclusions based on data mostly from thrombolysis or bare metal stents (BMS) era. Furthermore, contemporary treatment regimens have changed a lot over the last decade with common use of drug eluting stents (DES) and advances in adjunctive pharmacotherapy,but latest evidence is rare. The aim of this study is to evaluate the association between baseline platelet count (PC) and severe adverse outcomes following percutaneous coronary intervention (PCI) in current real-world practice both at short-term and long-term follow-up by analyzing data from the Beijing Heart and Metabolism Survey (BHMS).

2 Methods

2.1 Study design and patient population

BHMS is an investigator-initiated, multicenter cohort study conducted at five tertiary medical centers. The PC obtained at baseline, using a Coulter Counter method, was considered. From April 2004 to October 2010, a total of 21,620 consecutive patients receiving PCI were recruited.And only those implanted with DES were considered eligible for the study. To enhance homogeneity and ensure examination of a representative cohort in the context of contemporary treatment regimens, 843 (3.9%) patients receiving plain old balloon angioplasty without stent implantation and 1738 (8.0%) patients who underwent BMS implantation were excluded. Two hundred and eleven (1.0%) patients without completed baseline data and 40 (0.2%) with a terminal illness were also excluded. Thus a total of 18,788 patients constituted the cohort. Patients in our cohort were categorized as having low (＜ 150 × 1000/μL), normal (150–300 × 1000/μL), and high (≥ 300 × 1000/μL) baseline PC. In the overall cohort, the average length of in-hospital stay was 8 ± 6 days and the mean length of follow-up was about 2 years (25 ± 17 months). Figure 1 shows the study design and flow chart. The study protocol was reviewed and approved by the Ethics Committee of each participating institution.

Figure 1. The study flow chart. PCI: percutaneous coronary intervention; PC: platelet count; AHA/ACC: American Heart Association/American College of Cardiology; DES: drug eluting stent; BMS: bare metal stent.

2.2 Clinical endpoints

The primary endpoints were in-hospital and follow-up all-cause mortality. The secondary endpoint was major bleeding requiring a blood transfusion. The tertiary endpoints included length of stay before PCI, length of stay after PCI, length of in-hospital stay, and the total hospitalization expenses.

2.3 Statistical analysis

Statistical analysis was performed using Statistical Package for Social Sciences (SPSS). All categorical variables were expressed as percentages and compared with Pearson chi-square test or Fisher exact test; continuous variables were expressed as median (interquartile range) and nonparametric tests were used (Kruskal-Wallis test for ＞ 2 groups). Multivariate analyses with Cox proportional hazards methods derived the independent predictors of adverse events. Variables were selected for submission to the model if the univariate P value was ＜ 0.25 or the variable was of known clinical significance but failed to meet the critical α level for inclusion.[7–15]

3 Results

3.1 Baseline demographic and clinical characteristics

As an overall cohort, the mean ± SD age was 60 ± 11 years with the median age of 60 (52–68) years. Male patients (n = 13,922) account for 74.1% of the cohort. There were 5384 (28.7%), 11,425 (60.8%), 2832 (15.1%), and 2915 (15.5%) patients had a history of diabetes mellitus(DM), hypertension, hyperlipidemia and prior myocardial infaction (MI), respectively. A total of 13,283 patients had the myocardial dysfunction with different New York Heart Association (NYHA) functional classes, among which class II, III and IV accounted for 54.6%, 12.8% and 3.4%, respectively. The prevalences of stable coronary artery disease(SCAD), unstable angina pectoris (UAP) and acute myocardial infarction (AMI) were 23.2%, 48.5% and 28.3%,respectively.

3.2 Distribution of baseline PC

The distribution of baseline PC was a skewed distribution with a median value of 2 × 1000/μL (165–240 ×1000/μL) and a mean value of (206 ± 60) × 1000/μL (Figure 2-A). A majority of the baseline PC (14,633, 77.9%) were normally distributing in the range of 150–300 × 1000/μL(figure 2-B). In addition, there were 2884 (15.4%), 1271(6.8%) patients had their baseline PC ＜ 150 × 1000/μL, and≥ 300 × 1000/μL, respectively. Only 11 patients had their baseline PC lower than 50 × 1000/μL with the minimum value of 13 × 1000/μL (Figure 2-C). And 4 patients in the high group had their baseline PC higher than 600 × 1000/μL with the maximum value of 664 × 1000/μL (Figure 2-D).

Figure 2. The distribution of baseline platelet count. (A): The distribution of baseline platelet count in the overall cohort; (B): the distribution of baseline platelet count in the normal group; (C): the distribution of baseline platelet count in the low group; (D): the distribution of baseline platelet count in the high group.

3.3 Comparison of baseline characteristics among groups

As detailed in Table 1, among the 3 groups there were major differences in baseline clinical characteristics, which in the low group were almost the opposite of the high group.From an angiographic and procedural viewpoint (Table 2),patients with lower baseline PC were more likely to have left main (LM) disease and LMMVD. They were treated somewhat more frequently with single stent with shorter length and larger diameter. In addition, transradial approach PCI was more likely preferred in the low group. Compared with the normal group, patients in the high group were also more likely to have LM disease and were treated more frequently with single stent with relatively shorter length and larger diameter.

With respect to medications used in hospital and upon discharge (Table 3), there were no significant differencesregarding the use of clopidogrel among the 3 groups, whereas maintenance dose and loading dose of aspirin tended to be lower in the low group. The use of PPI was more frequent in the patients with lower baseline PC than the others.

Table 1. Baseline clinical characteristics.

Table 2. Angiographic and procedural characteristics.

Table 3. Medications used in hospital and upon discharge.

3.4 Association between baseline PC and clinical outcomes

3.4.1 Primary endpoints

In the overall cohort, 77 patients (0.4%) died in the hospital, and 120 patients (0.6%) died during the long-term follow-up (median 23.8 months). Compared with the normal group, both the low and high groups had the similar in-hospital and follow-up all-cause mortality (Table 4).In-hospital mortality rates for patients in the low, normal,and high group were 0.6%, 0.4%, and 0.4%, respectively; P= 0.259; and follow-up mortality rates for patients in the low, normal, and high group were 0.9%, 0.6%, and 0.7%,respectively; P = 0.079. After multivariable adjustment,patients with lower or higher baseline PC tended to have similar risks for both in-hospital and follow-up mortality compared with the normal group. As indicated in Table 5,hazard ratios (HRs) and 95% confidence intervals (CIs) of in-hospital death for the low and high group were 0.843(95% CI: 0.412–1.723; P = 0.639) and 0.668 (95% CI:0.236–1.890; P = 0.447), respectively; and the HRs of follow-up death for the low and high group were 1.204 (95%CI: 0.708–2.049; P = 0.493) and 0.942 (95% CI: 0.407–2.181; P = 0.889), respectively. Further subgroup analyses failed to demonstrate independent prognostic value of baseline PC in specific population groups except patients underwent transfemoral PCI (Figure 3). Kaplan–Meier curvesof in-hospital (Figure 4-A) and follow-up mortality (Figure 4-B) were presented in Figure 4. The cumulative survival rates in patients with low or high baseline PC continued to be similar to that in normal group (P = 0.548 and 0.082,respectively).

Table 4. Clinical outcomes at follow-ups.

Table 5. Independent predictors of in-hospital and long-term mortality.

Figure 3. Subgroup analyses of the prognostic value of baseline platelet count. (A): Relationship between baseline platelet count and in-hospital mortality in subgroups; (B): relationship between baseline platelet count and follow-up mortality in subgroups. AMI: acute myocardial infarction; CI: confidence intervals; HR: hazard ratio; SCAD: stable coronary artery disease; UAP: unstable angina pectoris.

Figure 4. Kaplan–Meier curves of in-hospital and follow-up mortality. (A): Kaplan–Meier curves of in-hospital mortality; (B): Kaplan–Meier curves of follow-up mortality.

3.4.2 Secondary endpoint

There was also no significant difference in the incidence of hemorrhage among groups (major bleeding requiring a blood transfusion in the low, normal, and high group were 0.5%, 0.3%, and 0.3%, respectively; P = 0.320). After multivariable adjustment, low (HR: 1.978; 95% CI: 0.975–3.818; P = 0.052) or high baseline PC (HR: 1.264; 95% CI:0.443–3.601; P = 0.662) did not significantly increase the risk of major bleeding (Table 6).

3.4.3 Tertiary endpoints

Although none of the tertiary endpoints were clinical adverse events, any of them could indirectly reflect the general incidence of severe adverse events. As indicated in Figure 5-A, there was no significant difference in the total hospitalization expenses among groups (P = 0.342). There were statistically significant differences in the length of in-hospital stay (median value in three groups were all 7 days, P ＜0.001) (Figure 5-B), the length of stay before PCI (median value in three groups were all 2 days, P = 0.047) (Figure 5-C), the length of stay after PCI (median value in 3 groups were all 4 days, P ＜ 0.001) (Figure 5-D), but the statistical differences did not translate into clinical importance.

Table 6. Independent predictors of major bleeding requiring a blood transfusion.

Figure 5. Comparisons of the tertiary endpoints among groups with different baseline platelet count. (A): Comparisons of the total hospitalization expenses among groups; (B): comparisons of the length of in-hospital stay among groups; (C): comparisons of the length of stay before PCI among groups; (D): comparisons of the length of stay after PCI among groups. CNY: ; PCI: percutaneous coronary intervention.

4 Discussion

The PCI technique and the adjunctive pharmacotherapy have made great progress in the following several decades.[16]However, thrombosis and hemorrhage have always been the the major cause of morbidity and mortality in patients underwent PCI.[1,2]Therefore, patients with impaired quantity and quality of platelets were often excluded from prospective randomized controlled trials because of the potential increased risks of thrombosis and hemorrhage following PCI. Few pooled post-hoc analyses[7–15]and cohort studies[13–15]drew inconsistent conclusions based on data mostly from thrombolysis or BMS era.

Gibson, et al.[12]demonstrated that in patients with ST-elevation myocardial infarction (STEMI) who were treated with aspirin, high baseline PC was independently associated with increased rates of reinfarction at 30-day follow-up.However, clopidogrel therapy abolished this increase in the risk of reinfarction as PC increased. Iijima, et al.[7]argued that in patients underwent PCI after pre-treatment with 600 mg clopidogrel, high baseline PC was still independently associated with 30-day mortality. Others[9–11]agreed with that, high baseline PC was independently associated with an increased risk of adverse events following PCI. Whereas there was one more post-hoc analysis[8]revealing that low baseline PC in STEMI patients underwent PCI was strongly associated with 30-day adverse events but not with any 2-year adverse events. Similarly, a cohort study[15]showed that in-hospital death rate was higher in patients with low baseline PC due to an increased mortality in AMI patients underwent urgent but not elective PCI. In another cohort study,[14]baseline PC was not an independent predictor of 30-day mortality in AMI patients after adjustment of confounders. Interestingly, a U-shaped association between baseline PC and long-term outcomes was also proposed.[13]Wu, et al.[17]demonstrated a significant association between baseline PC and clinical outcomes by meta-analysis of the above eight studies.[9–13,15–17]They confirmed a U-shaped relationship between baseline PC and the risk of mortality and adverse events. At 1-month follow-up, compared with the low PC group (＜ 150 × 1000/μL), the pooled relative risks of mortality and adverse events were 1.78 and 1.63 for the high PC group (＞ 350 × 1000/μL). At long-term followup, the pooled relative risks of mortality and adverse events were 1.48 and 1.28, respectively, for the high PC group.

However, the above studies have many limitations.Firstly, most of the patient population were AMI[9–12,14,16]or high-risk patients with acute coronary syndrome (ACS).[13,17]Even in the remainder one study,[15]the elective PCI accounted for less than 50%. Accordingly, conclusions drawn from the above studies could not be applicable to all CAD patients. Secondly, the sample sizes of such studies did not provide sufficient statistical power to detect low incidences of events in all prespecified groups according to clinical significance. Although equal interval classification[10,12,17]could increase statistical power, it might reduce the clinical significance of the cut-off points. Thirdly, not all patients in the above studies underwent PCI.[9,11–13,16,17]Lastly, and most important of all, some latest advances recommended by guidelines[18–20]were not reflected in the above studies,including DAPT, transradial approach for PCI, and PPI, etc.

Different from previous studies, we found that there were no significant differences among patients with varied baseline PC in severe adverse outcomes, including in-hospital mortality, long-term follow-up mortality, and major bleeding requiring a blood transfusion. Although the exact mechanism is not fully understood, several factors with well-established benefits may be involved in the changing pattern between baseline PC and outcomes, including increased use of clopidogrel added to aspirin,[21–24]transradial approach PCI,[25–27]PPI,[28,29]and optimization of stent implantation.[30,31]

5 Conclusions

There is no significant association between baseline PC and severe adverse outcomes following PCI in current real-world practice.

Acknowledgement

The Beijing Heart and Metabolism Survey (BHMS) was supported by grants from the Beijing Nova Program (No.Z121107002512053), the Beijing Health System High Level Health Technology Talent Cultivation Plan (No. 2013-3-013), the Beijing Outstanding Talent Training Program (No.2014000021223ZK32), the National Natural Science Foundation of China (No. 81100143), the Beijing Municipal Administration of Hospitals Clinical Medicine Development of Special Funding Support (No. ZYLX201303), and the National Key Clinical Specialty Construction Project.