Objective Whether lipoprotein(a) (Lp(a)) is a predictor for recurrent cardiovascular events (RCVEs) in patients with coronary artery disease (CAD) has not been established. This study, hence, aimed to examine the potential impact of Lp(a) on RCVEs in a real-world, large cohort of patients with the first cardiovascular event (CVE).
Methods In this multicentre, prospective study, 7562 patients with angiography-diagnosed CAD who had experienced a first CVE were consecutively enrolled. Lp(a) concentrations of all subjects were measured at admission and the participants were categorised according to Lp(a) tertiles. All patients were followed-up for the occurrence of RCVEs including cardiovascular death, non-fatal myocardial infarction and stroke.
Results During a mean follow-up of 61.45±19.57 months, 680 (9.0%) RCVEs occurred. The results showed that events group had significantly higher Lp(a) levels than non-events group (20.58 vs 14.95 mg/dL, p<0.001). Kaplan-Meier analysis indicated that Lp(a) tertile 2 (p=0.001) and tertile 3 (p<0.001) groups had significantly lower cumulative event-free survival rates compared with tertile 1 group. Moreover, multivariate Cox regression analysis further revealed that Lp(a) was independently associated with RCVEs risk (HR: 2.01, 95% CI: 1.44 to 2.80, p<0.001). Moreover, adding Lp(a) to the SMART risk score model led to a slight but significant improvement in C-statistic (∆C-statistic: 0.018 (95% CI: 0.011 to 0.034), p=0.002), net reclassification (6.8%, 95% CI: 0.5% to 10.9%, p=0.040) and integrated discrimination (0.3%, 95% CI: 0.1% to 0.7%, p<0.001).
Conclusions Circulating Lp(a) concentration was indeed a useful predictor for the risk of RCVEs in real-world treated patients with CAD, providing additional information concerning the future clinical application of Lp(a).
- Lipoproteins and hyperlipidaemia
- coronary artery disease
- cardiac risk factors and prevention
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Lipoprotein(a) (Lp(a)) is a low-density lipoprotein (LDL)-like particle consisting of an apolipoprotein B100 molecule covalently linked to a very large glycoprotein known as apolipoprotein(a).1 2 The specific physiological and vascular effects of this particle has not been fully understood. However, its structural homology to plasminogen and the most proatherogenic subtype may result in enhanced thrombogenic and atherogenic properties.2 Furthermore, it has been reported that Lp(a) can enter the arterial intima of humans,3 and in vitro and animal studies have shown that Lp(a) can promote inflammation, thrombosis and foam cell formation, most important mechanisms of atherosclerosis.4–6
In recent years, Lp(a) has been recognised as an most attractive and promising risk factor for cardiovascular disease (CVD).7 Although numerous studies including ours have demonstrated the important role of Lp(a) in promoting CVD in primary prevention8–12 and predicting cardiovascular events (CVEs) in secondary prevention,2 13–18 the prognostic value of Lp(a) for recurrent CVEs (RCVEs) risk in patients who have experienced CVEs has been controversial. As is well known, patients with previous CVEs are at high risk for ischaemic RCVEs, in spite of high-intensity statin therapy and other secondary prevention strategies.19 Further enhancement of risk stratification and high-risk patients identification in this patient population is urgent. We conducted this study to further evaluate the significance of Lp(a) in predicting the risk of RCVEs in patients with CAD who had experienced a first CVE.
Study design and population
As shown in figure 1, from March 2011 to December 2016, 11 906 Chinese patients scheduled for coronary angiography because of angina-like chest pain and/or positive treadmill exercise test and/or significant stenosis indicated by coronary CT angiography were consecutively recruited from four medical centres, including FuWai Hospital, XuanWu Hospital, AnZhen Hospital and the Fifth Hospital of Wuhan according to the same protocol. The blood samples for testing Lp(a) were sent to FuWai Hospital for unified measurement. Based on medical history, previous medical records, clinical manifestations, assistant examinations and coronary angiography data, 7846 patients with CAD who had experienced a first CVE (defined as myocardial infarction (MI), stroke, unstable angina, percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG)) between 2 months to 1 year before admission were evaluated. Next, 257 patients were excluded according to the criteria as described in our previous studies.20 21 All enrolled patients were prescribed secondary prevention therapy of CAD. Subsequently, 27 patients were lost to follow-up during the study. Thus, the resulting population consisted of 7562 patients with CAD with first CVEs. They were further divided into three groups according to Lp(a) tertiles. Details of the definition of CAD and exclusion criteria are mentioned in the online supplementary material.
Patient and public involvement
Patients or the public were not involved in the design, conduct, reporting or dissemination plans of our research. Details of the clinical assessment and biochemical analysis are mentioned in the online supplemental material.
Continuous variables are expressed as mean±SD or median (Q1–Q3 quartiles) as appropriate. The differences between groups were determined with the Student’s t-test, analysis of variance, Mann-Whitney U test or Kruskal-Wallis H test where appropriate. Categorical variables are presented as number (percentage) and analysed by χ2 test or Fisher’s exact test. The event-free survival rates among groups were calculated by the Kaplan-Meier method and compared by the log-rank test. Cox proportional hazard models were used to calculate the HRs and 95% CIs, and the predicting power of the model was calculated by receiver operating characteristic (ROC) curve analysis. Poisson regression analysis was performed to further demonstrate the association of Lp(a) with RCVEs. Restricted cubic spline (RCS) adjusted for age and sex was created to assess linearity assumptions of the relationship between Lp(a) and RCVEs. In addition, sensitivity and subgroup analyses were performed to better clarify the association of Lp(a) with RCVEs, which are mentioned in the online supplemental material. These analyses are made for the first subsequent event in participants with a prior history of CVE. To evaluate whether adding Lp(a) to the SMART risk score (detailed information is shown in online supplementary material),22 a prediction rule for recurrent vascular events, could improve the ability for predicting RCVEs, we calculated Harrell’s C-statistic, the continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI).23 24 In χ2 tests, for comparisons between any two of the three groups according to Lp(a) tertiles, two-tailed p values <0.0167 were considered statistically significant. For the other analyses, two tailed p values <0.05 were considered statistically significant. The statistical analyses were performed with SPSS V.24.0 software (SPSS, Chicago, Illinois, USA) and R language V.3.5.2 (Feather Spray).
Consistent with previous studies,15 17 Lp(a) levels had a skewed distribution with a tail towards the highest levels in the present study (online supplementary figure 1). The baseline characteristics of the entire study population as well as according to Lp(a) tertiles (<8.88, 8.88–26.44 and ≥26.45 mg/dL) are detailed in table 1. Patients with lower Lp(a) levels were more likely to be males, smokers and hypertensive, while those in tertile 3 of Lp(a) had lower diastolic blood pressure, glycosylated haemoglobin (HbA1c) and diabetes prevalence. There was an ascending gradient regarding the proportion of baseline P2Y12 inhibitor, statin and β-blocker uses, as well as concentrations of total cholesterol, high-density lipoprotein cholesterol, LDL cholesterol (LDL-C), apolipoprotein B, high-sensitivity C reactive protein (hsCRP), fibrinogen and erythrocyte sedimentation rate (ESR) across Lp(a) tertiles.
Lp(a) levels and RCVEs
Over an average of 61.45±19.57 months follow-up, 680 RCVEs (251 cardiovascular deaths, 165 non-fatal MIs and 264 strokes) were recorded, representing 17.5 events per 1000 person-years. As shown in online supplementary table 1, patients with RCVEs were slightly older, and had lower left ventricular ejection fraction and a lower proportion of revascularisation (included previous and current PCI/CABG) compared with those in non-events group (both p<0.05). Additionally, the events group had a higher proportion of patients with hypertension and diabetes, and higher levels of HbA1c, Lp(a), hsCRP, fibrinogen and ESR compared with non-events group (all p<0.05). However, the proportion of statin or high-intensity statin use between two groups had no significant difference (both p>0.05).
The corresponding RCVEs per 1000 person-years were 12.9 (95% CI: 8.6 to 17.2), 19.0 (95% CI: 13.7 to 24.3) and 20.6 (95% CI: 15.3 to 26.7), respectively, from Lp(a) tertile 1 to tertile 3. The prevalence of total RCVEs and cardiovascular deaths in tertile 2 and 3 groups were significantly higher compared with tertile 1 group (all p<0.0167). Meanwhile, the incidence of recurrent non-fatal MI and stroke in tertile 3 of Lp(a) group were also significantly higher than tertile 1 group, respectively (both p<0.0167; online supplementary figure 2). Furthermore, the Kaplan-Meier analysis showed that subjects in Lp(a) tertile 2 and 3 had significantly lower total event-free survival rates compared with those in Lp(a) tertile 1 (both p<0.01, online supplementary figure 3). When cardiovascular death and stroke were considered independently, we observed that Lp(a) tertile 2 and 3 groups also had significantly lower event-free survival rates than tertile 1 group (all p<0.05, online supplementary figure 3). Meanwhile, subjects in tertile 3 of Lp(a) group had greater risk for non-fatal MI compared with those in tertile 1 group (p=0.011, online supplementary figure 3).
The adjusted HRs, 95% CI and survival curves of RCVEs according to Lp(a) levels are shown in table 2 and figure 2. Subjects in Lp(a) tertile 2 had a 1.60-fold higher risk, while those in tertile 3 had a 1.72-fold higher risk of RCVEs occurrence in the crude model compared with the reference group (Lp(a) tertile 1). Additional adjustment for other potential covariates did not change this association (HR: 1.69; 95% CI: 1.21 to 2.36; HR: 2.01; 95% CI: 1.44 to 2.80; respectively), the mean area under the curve (AUC) for this model was 0.698 (95% CI: 0.664 to 0.731, p<0.001; online supplementary figure 4). The Poisson regression analysis also showed that both tertile 2 [relative risk (RR): 1.60, 95% CI: 1.17 to 2.19, p=0.003] and tertile 3 (RR: 1.93, 95% CI: 1.42 to 2.63, p<0.001) of Lp(a) had significant and independent association with RCVEs (online supplementary table 2). Moreover, as shown in online supplementary figure 5, RCS showed a strong trend towards non-linear positive association of Lp(a) with RCVEs. The significant association between Lp(a) and RCVEs remained unchanged in a sensitivity analysis in which each of the other significant variables in univariate analysis was forced into the model with continuous Lp(a) (per 1−SD increase) (table 3). When the RCVEs were considered separately, we observed that tertile 3 of Lp(a) group remained significantly associated with a 1.98-fold (95% CI: 1.34 to 2.92) higher risk of cardiovascular death, a 1.57-fold (95% CI: 1.01 to 2.46) higher risk of non-fatal MI and a 1.98-fold (95% CI: 1.18 to 3.32) higher risk of stroke (p<0.05, respectively). In addition, further subgroup analysis suggested that for patients with a first acute coronary syndrome (ACS) event, elevated Lp(a) levels had a significant association with RCVEs (HR: 1.72, 95% CI: 1.30 to 2.27, p<0.001 for Lp(a) tertile 2 vs tertile 1; HR: 1.95, 95% CI: 1.48 to 2.58, p<0.001 for Lp(a) tertile 3 vs tertile 1), and that for subjects with a previous stroke event, Lp(a) tertile 3 group had a much higher risk for RCVEs compared with tertile 1 group (HR: 2.46, 95% CI: 1.19 to 5.11, p=0.015; online supplementary table 3). The ROC curve analysis showed that the AUC for the level of Lp(a) was higher in patients with a previous stroke (AUC: 0.63, 95% CI: 0.55 to 0.71) compared with those with a previous ACS (AUC: 0.55, 95% CI: 0.52 to 0.58). Another subgroup analysis according to sex showed that men in Lp(a) tertile 2 had a 1.92-fold (95% CI: 1.28 to 2.85, p=0.001) higher risk while those in tertile 3 had a 2.25-fold (95% CI: 1.51 to 3.34, p<0.001) higher risk for RCVEs compared with men in Lp(a) tertile 1. However, we observed no significant differences of RCVEs risk among Lp(a) tertiles in women (online supplementary table 4).
Finally, we assessed whether the addition of Lp(a) to established cardiovascular risk factors improves risk stratification for RCVEs. The C-statistic value of SMART risk score model,22 which estimates the 10-year risk for recurrent vascular events in patients with manifest CVD, was 0.682 (95% CI: 0.646 to 0.718). The addition of Lp(a) tertile to the model showed a significant improvement in C-statistic (∆C-statistic: 0.018 (95% CI: 0.011 to 0.034), p=0.002), and also a significant increase in NRI (6.8%, 95% CI: 0.5% to 10.9%, p=0.040) and IDI (0.3%, 95% CI: 0.1% to 0.7%, p<0.001).
Even though circulating Lp(a) has become a novel biomarker for cardiovascular risk prediction, the potential role of Lp(a) in predicting the risk of RCVEs has not been determined. Notably, in this study on a sizeable population with CAD who had experienced a first CVE and received secondary prevention treatment, we clearly demonstrated that baseline Lp(a) levels were significantly associated with future RCVEs. This relation persisted even after adjustment for established cardiovascular risk factors, suggesting that circulating Lp(a) is really a useful biomarker for predicting RCVEs. Undoubtedly, current evidence may provide further information for understanding the clinical significance of Lp(a) in patients with CAD.
The risk of RCVEs has been increased in patients with established CAD.25 26 Despite ongoing treatment with statin therapy, patients with stable CAD remain at substantial risk of RCVEs. Notably, for patients with CAD with a previous CVE, such as ACS, the incidence of RCVEs is even higher. Among 46 694 initially stabilised patients enrolled in four post-ACS trials, 9.2% of them experienced a RCVE during a median follow-up of 1 year.27 In another study of survivors of first MI, the incidence of recurrence were 7.2% for women and 5.6% for men at 1 year, and 16.2% and 13.9%, respectively, at 7 years.28 Meanwhile, for patients surviving a stroke, all-cause death rate was 24.5% at 1 year, while 41.3% at 4 years.29 However, there is very substantial variation in the risk of RCVEs.30 With all modifiable risk factors at guideline-recommended targets under optimal treatment, half of the patients would have a 10-year risk of RCVEs <10%, while some other patients will remain at >20% and even >30% 10-year risk.30 Therefore, despite the traditional risk factors, the exploration of new biomarkers predicting RCVEs is significant and necessary.
As is well known, there is considerable evidence that a high level of Lp(a) is an independent risk factor for incident CVD.1 8 9 31–34 Moreover, for patients with established CAD, an association between Lp(a) and cardiovascular outcomes has also been well confirmed.2 13–18 However, when it comes to the association between Lp(a) and RCVEs, there have been few studies, of which the results are inconsistent. The disparity among the existed studies might be associated with population selection, clinical phenotype of CAD, different kit of Lp(a) measurement, sample size, follow-up duration and RCVEs type. For instance, in the early Scandinavian Simvastatin Survival Study, including 4444 patients with CAD with a history of either acute MI or angina pectoris, Lp(a) was found to be a significant predictor of recurrent coronary events and all-cause death.35 A study on 215 obese postinfarction patients with 26 months follow-up indicated that elevated Lp(a) was a highly significant risk marker of recurrent cardiac events (cardiac death, MI or unstable angina).36 Besides, in the study by Zhou et al, with 713 patients with ACS and 1–2 years follow-up, Lp(a) levels were showed to be an independent risk factor for RCVEs including definite or suspected cardiac death, non-fatal MI, coronary revascularisation and stroke.37 Nevertheless, a case-control study (n=520) with 22 months follow-up found that baseline Lp(a) was not an independent risk factor for RCVEs (cardiac death, MI and target lesion revascularisation).38 In a recent study, for patients with recent ACS, Lp(a) concentration was also found to be not associated with subsequent risk of major ischaemic CVEs, including CAD death, non-fatal coronary event and ischaemic stroke.19 Hence, the association of circulating Lp(a) with RCVEs may need further investigation.
In the present study, with a large sample size and long-term follow-up, we indeed found that patients with RCVEs had higher Lp(a) levels compared with those without recurrent events. Moreover, elevated Lp(a) (higher two tertiles) was significantly and independently associated with higher risk of RCVEs including cardiovascular death, non-fatal MI and stroke in patients with CAD who had experienced a first CVE. The subsequent sensitivity analysis further demonstrated this association. In subgroup analysis, tertile 3 of Lp(a) remained associated with a significantly higher risk of cardiovascular death, non-fatal MI and stroke, respectively. The relative lower HR for non-fatal MI may be due to the allocation of fatal-MI to cardiovascular death. For patients with either a first ACS or stroke event, tertile 3 of Lp(a) was both significantly associated with RCVEs. However, the HR for tertile 3 of Lp(a) and AUC for continuous Lp(a) were much higher in patients with a first stroke compared with those with a previous ACS, suggesting that Lp(a) might be differentially more associated with RCVEs risk in patients with previous cerebrovascular issues. Furthermore, another subgroup analysis according to sex showed that the predicting role of Lp(a) for RCVEs mainly existed in men, which may be attributed to the large proportion (73.9%) of our subjects were men. Future studies on a large sample of women may be needed to further clarify the specific association between Lp(a) and RCVEs. Moreover, C-statistic, NRI and IDI analyses further demonstrated that Lp(a) could slightly but significantly improve RCVEs risk prediction on the basis of the established SMART Risk Score model.22 Most prior analyses suggested that the relation between Lp(a) and cardiovascular risk inflects at a concentration of about 30 mg/dL.2 39 Nevertheless, in this study, it is surprising that patients in the middle tertile of Lp(a) (centroid 15 mg/dL) also had a higher risk of RCVEs compared with subjects in the lowest tertile (centroid 5 mg/dL), which may be attributed to the distinguished Lp(a) concentrations among different ethnicities.40 It has been reported that Lp(a) levels are lowest in non-Hispanic Caucasians (median: 12 mg/dL; IQR: 5–32), Chinese (11, 4–22) and Japanese (13, 5–26), slightly higher in Hispanics (19, 8–43) and even higher in blacks (39, 19–69).40
It has been well known that statins could improve clinical prognosis in patients with CAD.41 Moreover, the current guidelines recommend the use of statins, especially high-intensity statins in patients with established CAD.42 43 Indeed, previous studies suggested that the higher intensity statin therapy showed better clinical outcomes over the lower intensity statin therapy, which mainly focusing on Western populations.44 However, East Asian populations may have lower baseline LDL-C levels, better statin responsiveness and greater vulnerability to the side effects of statin compared with Western populations.45 Additionally, a recent study failed to demonstrate the enhanced clinical efficacy of high-intensity statin therapy in Chinese population.46 Accordingly, our data showed a low proportion of high-intensity statin use and the non-events group had a slight but not significant higher use of statins and high-intensity statins at baseline compared with events group.
Our findings are limited by several factors. First, as inherent to the nature of observational and prospective study, our results are subject to confounding factors and the baseline risk factors may change during the follow-up. Second, we did not have the follow-up data of Lp(a), which may improve the significance of an association between Lp(a) and CVD outcomes. However, Lp(a) is an inherited lipoprotein, which cannot be altered by diet or exercise and may be relatively stable in life. Third, Lp(a) was measured by immunoturbidimetry method, which was not apolipoprotein(a) isoform independent. However, an Lp(a) protein validated standard was used to calibrate the examination, making the assay relatively isoform independent. Fourth, the information about anticoagulation therapy was not recorded, which might have an effect on the outcomes. Fifth, we lacked the accurate records of the duration of P2Y12 inhibitors before admission, which may have a significant impact on RCVEs. However, all enrolled subjects were prescribed secondary prevention therapy and P2Y12 inhibitors were used if necessary.
In summary, the present study, featured in real-world manner with a large sample size and long-term follow-up, gave further evidence that Lp(a) is independently associated with the risk of RCVEs in patients with CAD under secondary prevention therapy, providing additional information for the clinical application of Lp(a).
What is already known about this subject?
Previous studies have demonstrated the important role of lipoprotein(a) (Lp(a)) in promoting cardiovascular disease in primary prevention and predicting cardiovascular events in secondary prevention.
The prognostic value of Lp(a) for recurrent cardiovascular events risk in patients who have experienced a first cardiovascular event has been controversial.
What does this study add?
In this large sample size and long-term follow-up study, with real-world practice, our data suggested the significance of Lp(a) in predicting recurrent events in patients with coronary artery disease experienced a previous cardiovascular event.
How might this impact on clinical practice?
Circulating Lp(a) concentration may be treated as a useful predictor for the risk of recurrent cardiovascular events in real-world treated patients with coronary artery disease.
Our findings provide additional information concerning the future clinical application of Lp(a).
The authors would like to thank all the staff and participants of this study for their important contributions.
Contributors H-HL designed this study and wrote the manuscript; Y-XC, J-LJ, H-WZ were in charge of the statistical analysis and interpretation of the results; QH, Y-FL, Y-LG, C-GZ, N-QW, YG, R-XX and L-FH conducted this study and collected data. J-JL designed this study and made critical revisions of the manuscript.
Funding This work was partially supported by the Capital Health Development Fund (201614035) and CAMS Major Collaborative Innovation Project (2016-I2M-1-011) awarded to J-JL, the Fundamental Research Funds for the Central Universities (2019-XHQN09) and the Youth Research Fund of Peking Union Medical College (2019-F11) awarded to H-HL and the Fundamental Research Funds for the Central Universities (2018-XHQN03) and the Youth Research Fund of Peking Union Medical College (2018-F05) awarded to YG.
Disclaimer The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The research protocol complied with the Declaration of Helsinki and was approved by the hospital’s ethical review board (Fu Wai Hospital & National Center for Cardiovascular Diseases, Beijing, China). Each subject provided written, informed consent before enrolment.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request from the corresponding author Professor J-JL.