WHAT IS ALREADY KNOWN ON THIS TOPIC

Previous studies have shown that percutaneous coronary intervention (PCI) improves symptoms, eliminates ischaemia and reduces the need for urgent revascularisations. In the modern era of PCI, no study has shown any mortality difference between PCI and medical therapy.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

PCI should be considered in addition to medical therapies when assessing symptomatic patients with chronic coronary syndrome and obstructive coronary artery disease. Patients with chronic coronary syndrome should be made aware of available treatment strategies including the benefits after revascularisation with PCI.

Introduction

Lifestyle modifications and pharmacological therapy play a crucial role in managing symptoms and halting atherosclerotic progression in individuals with chronic coronary syndrome (CCS).1–3 Percutaneous coronary intervention (PCI) improves outcomes in acute coronary syndrome but its role in CCS remains controversial.4–6 Several randomised studies have attempted to evaluate the efficacy of revascularisation versus medical therapy (MT) among patients with CCS. In addition to being inconclusive, most trials were conducted before the widespread use of risk-modifying therapies with statins, or before the advent of drug-eluting stents. Randomised clinical trials evolved from selecting patients based on quantitative coronary artery disease (CAD) assessment to encompassing patients exhibiting ischaemia, flow-limiting lesions, viable myocardium and high-risk profile, aiming to pinpoint patients who might benefit from revascularisation. Nevertheless, no discernible enhancement in survival following PCI-based revascularisation has been proven.7–11 Despite the lack of survival benefit, patients with CCS typically comprise one-third of all patients undergoing coronary angiography.12–14 The widespread adoption of PCI for CCS can be explained by the challenges associated with extrapolating clinical trial outcomes to real-world scenarios. These challenges arise from disparities in study design, patient selection criteria and the reliance on composite endpoints, often demonstrating reductions in unplanned revascularisation and thereby favouring PCI over MT alone. Consequently, this trend has sparked increased interest in understanding the significance of a therapy beyond mortality. Hierarchical statistical analyses have been developed to address this issue.15 16 These statistical methods prioritise pivotal endpoints, such as mortality, emphasising their significance when assessing a composite outcome. This shift aims to refine the understanding of treatment impact when evaluating therapeutic efficacy without compromising statistical power.

The primary objective of the present study was to evaluate the potential benefit of PCI in patients on guideline-recommended MT compared with guideline-recommended MT alone. Using nationwide, real-world data, we adopted a hierarchical statistical approach to provide insights into the clinical outcomes and prognostic implications of PCI in patients with CCS.

MethodsData source

This is a nationwide registry-based study conducted using the Swedish Coronary Angiography and Angioplasty Registry (SCAAR)17 and the Swedish Prescribed Drugs Registry. SCAAR collects information from all patients in Sweden undergoing coronary angiography. The registry includes extensive information on patient characteristics, angiographic findings and procedural information. The Swedish Prescribed Drugs Registry contains data on all MTs dispensed in Sweden since 2005. SCAAR and the national prescribed drug registries were linked to the National Population Registry, which includes all International Classification of Diseases (ICD) codes from all admissions in Sweden since 1987.

Study design, study population and outcome

A flow chart depicting inclusion and exclusion criteria is presented in figure 1. For this study, SCAAR was used to identify all patients in Sweden with CCS undergoing coronary angiography between 2010 and 2020. The first angiography during the study period was considered the inclusion date and all patients were included only once. Guideline-recommended MT was defined in accordance with class 1 recommendations from the European Society of Cardiology guidelines (table 1).18 All included patients were on lipid-lowering therapy with statins and at least one anti-ischaemic drug (beta blocker, calcium channel inhibitor and/or long-acting nitrate). Based on comorbidities, ACE inhibitor, angiotensin receptor blocker, beta blocker, aspirin and clopidogrel were also included in the definition of guideline-recommended MT. The Swedish Prescribed Drugs Registry was used to collect data on MTs and patients without dispensed guideline-recommended MT 180 days prior to 7 days after inclusion were excluded. Asymptomatic patients, patients with unknown symptoms, patients with no significant coronary stenosis and patients referred to coronary artery bypass graft surgery or valvular surgery were also excluded. Eligible patients were stratified into two groups according to treatment strategy. The first group, the PCI group, consisted of patients undergoing ad hoc PCI or referred for PCI within 45 days following their angiogram. The second group included patients receiving MT only.

Figure 1

Flow chart. All patients with chronic coronary syndrome undergoing angiography in Sweden between 2010 and 2020 were included and stratified into two groups according to treatment strategy: MT alone versus PCI and MT. One-to-one propensity score (PS) matching using a calliper method was carried out based on clinically relevant baseline variables. The propensity-matched study population included 15 440 patients. *A significant stenosis was defined as a luminal stenosis with ≥50% narrowing of the luminal diameter. CABG, coronary artery bypass graft surgery; MT, medical therapy; OMT, optimal medical therapy; PCI, percutaneous coronary intervention.

Table 1

Guideline-recommended medical therapy

Outcomes

The primary outcome was net adverse clinical events (NACE) within a 5-year follow-up. NACE was defined as the first occurrence of all-cause death, myocardial infarction (MI), bleeding or urgent revascularisation. Secondary outcomes included the individual components of NACE, major adverse cardiovascular events (MACE) and cardiovascular mortality. MACE was defined as first occurrence of all-cause death, MI or urgent revascularisation. Urgent revascularisation was defined as a revascularisation with PCI due to acute coronary syndrome or revascularisation with coronary artery bypass graft surgery. MI was defined as a new discharge diagnosis of MI (ICD-10: I21–I22), including periprocedural MI. Stroke was used as a safety outcome. A safety outcome refers to any outcome measure that pertains to the safety of the intervention being studied. Censorship dates and death status were ascertained by the National Board of Health and Welfare by deterministic linkage to the National Population Registry. Information on dates of MI, stroke and bleeding was obtained through similar linkage to the National Patient Registry and were defined according to ICD codes which are presented in online supplemental table 1. Follow-up was calculated from the inclusion angiography and all outcomes were ascertained up to 15 January 2022, with complete follow-up available for all patients. A long-term analysis was also conducted with 10-year follow-up.

Statistical analysis

Normal distribution was verified for all continuous variables and presented as means with SD and assessed using independent t-test. Categorical data are presented as counts with percentages and differences between categorical variables were studied using the χ2 test. Logistic regression was used to estimate propensity scores (PS) including the following variables: inclusion date, sex, age, smoking status, diabetes, stroke, kidney disease, heart failure, hypertension, hyperlipidaemia, previous coronary artery bypass graft surgery, previous PCI, disease extent on angiogram, Canadian Cardiovascular Society score, use of fractional flow reserve or instantaneous wave-free ratio, lesion ≥70% on angiography and country of birth. The variables were carefully selected a priori and the proportion of missing values in variables of interest was low (online supplemental table 2). One-to-one PS matching was then used with the nearest-neighbour approach using a calliper method, with calliper 0.001. Outcome was assessed with matched win ratio using the PS-matched pairs.15 In the matched win ratio analysis of NACE, all-cause mortality was ranked highest, followed by MI, bleeding and urgent revascularisation. For each pair, all-cause mortality was analysed first. If both patients died, the patient with the longest survival time won. If one of the patients died, the other patient must be followed for longer in order to win. If neither died, MI was analysed, followed by bleeding and, lastly, urgent revascularisation. If no win was assigned after analysing urgent revascularisation, the comparison ended in a tie. For MACE, all-cause mortality was ranked highest, followed by MI and urgent revascularisation. The components of NACE were also individually analysed with win ratio. The win ratio is presented along with p value and 95% CI. In addition to win ratio, traditional statistical methods, that is, Kaplan-Meier estimator and Cox proportional hazards models, were used and presented as event rates along with HR with 95% CI and p value. All Cox proportional hazards models used were univariable using the PS-matched cohort. Subgroup analysis was carried out for sex, age (80 years or older vs younger than 80 years), diabetes, hypertension, previous revascularisation, disease extent on angiogram (one-vessel disease vs multivessel disease and/or left main disease), Canadian Cardiovascular Society score grading (Canadian Cardiovascular Society score I vs II–IV), beta blockers, calcium channel inhibitors, long-acting nitrates, use of fractional flow reserve/instantaneous wave-free ratio and lesions ≥70% (patients with at least one lesion ≥70% vs patients with no lesion ≥70%). The subgroups were analysed on NACE using unmatched win ratio and Cox regression models, with p value for interaction. An additional supplementary subgroup analysis was conducted, focusing on all-cause mortality. A sensitivity analysis was conducted for the study population before PS matching using inverse probability weighting, including the variables used for the PS matching. All analyses were conducted on complete case data. A two-sided p value <0.05 was considered statistically significant. The unmatched win ratio analyses were conducted in R (V.4.2.2; The R Foundation for Statistical Computing, Vienna, Austria; package WinRatio). Data management and all other statistical analyses were done in STATA SE (V.17.0; StataCorp, Texas).

ResultsStudy population characteristics

After PS matching, two groups of 7220 patients were formed. The baseline characteristics were similar across the two groups, and there were no significant differences in the variables used in the PS model (table 2 and online supplemental figure 1). There was no difference in PS between the two groups (p=0.686) (online supplemental table 3). The mean age was 68.3 years, 77.8% of the patients were males, 77.0% had symptoms corresponding to Canadian Cardiovascular Society scores I–II and 62.7% had one to two-vessel disease on angiogram (table 2).

Table 2

Baseline characteristics after PS matching

Five-year follow-up

In the hierarchical win ratio analysis of NACE, PCI+MT was associated with significantly better outcome when compared with MT alone (matched win ratio: 1.28 (95% CI 1.20 to 1.36, p<0.001) and HR: 0.73 (95% CI 0.69 to 0.77, p<0.001)) (table 3, figures 2 and 3). PCI+MT was also associated with significantly lower rates of MACE (matched win ratio: 1.38 (95% CI 1.29 to 1.48, p<0.001) and HR: 0.69 (95% CI 0.65 to 0.73, p<0.001)) (table 3 and figure 3). When analysing the components of NACE individually, significantly more wins were observed for PCI+MT when analysing MI (matched win ratio: 1.15 (95% CI 1.04 to 1.28, p=0.008) and HR: 0.89 (95% CI 0.81 to 0.98, p=0.016)). PCI+MT was also associated with lower rates of urgent revascularisation (matched win ratio: 1.85 (95% CI 1.69 to 2.03, p<0.001) and HR: 0.58 (95% CI 0.54 to 0.63, p<0.001)) and cardiovascular mortality (matched win ratio: 1.15 (95% CI 1.00 to 1.34, p=0.044) and HR: 0.87 (95% CI 0.76 to 0.99, p=0.039)). No significant difference was observed for all-cause mortality (matched win ratio: 1.10 (95% CI 0.99 to 1.21, p=0.073) and HR: 0.92 (95% CI 0.84 to 1.00, p=0.060)), bleeding (matched win ratio: 0.96 (95% CI 0.85 to 1.08, p=0.502) and HR: 1.02 (95% CI 0.92 to 1.14, p=0.680)) or stroke (matched win ratio: 1.01 (95% CI 0.86 to 1.20, p=0.867) and HR: 1.01 (95% CI 0.87 to 1.18, p=0.859)) (table 3, figures 3 and 4 and online supplemental figure 2). The sensitivity analysis using inverse probability weighting was in line with the main results (online supplemental table 4).

Figure 2

Main results. Matched win ratio analysis of net adverse clinical events (NACE). NACE was defined as (1) all-cause mortality, (2) myocardial infarction, (3) bleeding or (4) urgent revascularisation within 5 years. When calculating the win ratio, the components of NACE were ranked in this order. Depicted percentages are percentages of wins, losses and ties. The win ratio is presented together with 95% CI and p value. MT, medical therapy; PCI, percutaneous coronary intervention.

Figure 3

Outcomes. Time-to-event Kaplan-Meier curves illustrating the event rate of (A) NACE, (B) MACE, (C) all-cause mortality, (D) myocardial infarction, (E) bleeding and (F) urgent revascularisation. MACE, major adverse cardiovascular event; MT, medical therapy; NACE, net adverse clinical event; PCI, percutaneous coronary intervention.

Figure 4

Cardiovascular mortality. Time-to-event Kaplan-Meier curves illustrating the event rate of cardiovascular mortality. MT, medical therapy; PCI, percutaneous coronary intervention.

Subgroup analysis

Significant PCI-by-subgroup interactions with better outcome after PCI were observed for patients younger than 80 years (interaction p value, <0.001), no diabetes (interaction p value, 0.001), no hypertension (interaction p value, 0.007), no previous revascularisation (interaction p value, <0.001), multivessel and/or left main disease (interaction p value, 0.002), no calcium channel inhibitor (interaction p value, 0.038), no long-acting nitrate (interaction p value, <0.001), use of fractional flow reserve/instantaneous wave-free ratio (interaction p value, <0.001) and patients with at least one lesion ≥70% (interaction p value, <0.001) (figure 5). In a supplementary subgroup analysis investigating all-cause mortality, PCI+MT was associated with significantly better survival when used for patients with a ≥70% lesion. Simultaneously, MT alone was associated with significantly lower rate of all-cause mortality when used for patients with no ≥70% lesion (interaction p value, <0.001) (online supplemental figure 3).

Figure 5

Subgroup analysis. Forest plot for subgroup analysis. Win ratio is a statistical method used for composite outcomes and allows sorting the components of the composite by clinical importance. *The subgroups were analysed on net adverse clinical events (NACE) using unmatched win ratio, matched win ratio is not possible with uneven pairs. CSS, Canadian Cardiovascular Society score grading of angina; FFR, fractional flow reserve; iFR, instantaneous wave-free ratio; LM, left main; MT, medical therapy; PCI, percutaneous coronary intervention.

Ten-year follow-up

In the 10-year follow-up analysis, PCI+MT was associated with lower rates of NACE (matched win ratio: 1.23 (95% CI 1.16 to 1.31, p<0.001) and HR: 0.79 (95% CI 0.75 to 0.82, p<0.001)), MACE (matched win ratio: 1.31 (95% CI 1.23 to 1.40, p<0.001) and HR: 0.75 (95% CI 0.71 to 0.79, p<0.001)), all-cause mortality (matched win ratio: 1.10 (95% CI 1.01 to 1.20, p=0.032) and HR: 0.92 (95% CI 0.86 to 0.99, p=0.022)), cardiovascular mortality (matched win ratio: 1.17 (95% CI 1.03 to 1.33, p=0.012) and HR: 0.88 (95% CI 0.79 to 0.98, p=0.020)) and urgent revascularisation (matched win ratio: 1.82 (95% CI 1.67 to 2.00, p<0.001) and HR: 0.61 (95% CI 0.56 to 0.66, p<0.001)). For MI, PCI+MT was associated with significantly more wins but not a significantly lower HR (matched win ratio: 1.12 (95% CI 1.02 to 1.23, p=0.022) and HR: 0.96 (95% CI 0.88 to 1.04, p=0.286)). No significant difference was observed for bleeding (matched win ratio: 0.95 (95% CI 0.85 to 1.06, p=0.379) and HR: 1.05 (95% CI 0.95 to 1.15, p=0.322)) and stroke (matched win ratio: 0.97 (95% CI 0.84 to 1.12, p=0.680) and HR: 1.05 (95% CI 0.93 to 1.19, p=0.394)) (table 3, figures 3 and 4 and online supplemental figure 2).

Discussion

Among patients with CCS contemporarily treated with guideline-recommended MT according to the European Society of Cardiology class I recommendations and undergoing coronary angiography, a strategy of revascularisation with PCI plus guideline-recommended MT demonstrated a prognostic benefit compared with a conservative strategy of MT only. The results, determined through a comprehensive hierarchical analysis, were driven by a higher rate of MI and unplanned revascularisation in the MT group. The use of PCI+MT was also associated with a lower risk of cardiovascular death at 5 and 10 years and all-cause mortality at 10 years.

Five randomised clinical trials have been conducted comparing the prognostic benefit of PCI to a conservative strategy of MT in the era of modern PCI and risk-modifying therapy (BARI-2D, COURAGE, ISCHEMIA, FAME-2 and REVIVED-BCIS). None of these trials have demonstrated any survival benefit associated with PCI when compared with MT alone. This study sought to evaluate the intermediate and long-term outcome of symptomatic patients with CCS exhibiting obstructive CAD and treated with antianginal therapy. A core set of inclusion and exclusion criteria aimed at mitigating selection bias and confounding were applied. These criteria excluded the frailest patients not deemed suitable candidates for an invasive strategy and resulted in the inclusion of patients who theoretically would benefit of revascularisation, that is, patients with obstructive CAD having symptoms despite guideline-recommended medical treatment. By employing a hierarchical statistical approach, we were also able to address the reliance on composite outcomes. This study provides real-world descriptive data on patients with CCS undergoing coronary angiography and their outcomes. Using a large sample size and extensive follow-up time it reassures the long-term safety and efficacy of PCI compared with contemporary guideline-recommended MT alone. Our results align with available studies even considering the demographic differences between real-world patients and those enrolled in clinical trials. Notably, in the FAME-2 study, revascularisation resulted in a significant reduction in the primary composite endpoint, which included death, non-fatal MI and urgent revascularisation.19 The absolute risk reduction in MACE after a 5-year period in our study was 8.5%, compared with 13.1% in the FAME-2 trial.19 While the rate of urgent revascularisation in the PCI group was significantly lower in the FAME-2 study compared with ours (6.3% vs 13.3%), the rate of urgent revascularisation in the MT group was strikingly similar (20.5% vs 21.1%). This difference could be explained by the focus on achieving complete revascularisation of flow-limiting lesions. Like the FAME-2 study, ISCHEMIA did not demonstrate any advantage of an early invasive strategy over a conservative approach during the 5-year follow-up period.9 Consistent with previous studies, our study showed no difference in all-cause death between groups at the intermediate 5-year follow-up. Both the ISCHEMIA trial and FAME-2 showed a trend towards reduced rates of cardiovascular or cardiac-related deaths after revascularisation/invasive strategy, with HR of 0.87 (95% CI 0.66 to 1.15) and HR of 0.70 (95% CI 0.47 to 1.04), respectively. In contrast, our results show that PCI is associated with lower rates of cardiovascular death, with similar effect estimates to those observed in ISCHEMIA and FAME-2 (table 3 and figure 4). The longer follow-up time and increased sample size in our study potentially facilitated the identification of a modest yet statistically significant difference in all-cause mortality between groups, favouring the PCI group at the 10-year mark. A slightly larger effect estimate (HR: 0.86, 95% CI 0.73 to 0.97, p=0.02) was observed in the 10-year follow-up of the STITCH-2 study which investigated revascularisation with coronary artery bypass graft surgery versus MT in patients with ischaemic cardiomyopathy.20 Like the STITCH-2 study, the REVIVED-BCIS investigation did not demonstrate any prognostic advantage with revascularisation using PCI over MT in patients with ischaemic cardiomyopathy in its original protocol.9 Similarly, the BARI-2D study showed no benefit with PCI compared with MT in patients with diabetes mellitus.21 In our study, PCI appears to confer more substantial benefits to patients without comorbidities. This trend might be explained by the notion that the impact of CCS on outcomes is less pronounced in individuals with concurrent conditions like heart failure or diabetes. This is supported by the subgroup analysis of all-cause mortality, in which patients with lesions exhibiting ≥70% luminal stenosis emerged as the sole disease modifier associated with lower risk of all-cause mortality when PCI was used. Contrary, for patients without lesions exhibiting ≥70% luminal stenosis, the risk of all-cause mortality was lower when a conservative treatment strategy of MT alone was used.

Limitations

This study has several limitations. First and foremost, underlying selection bias and confounding cannot be ruled out and the SCAAR registry lacks information on why a particular patient was selected for PCI+MT or MT alone. Decisions may be valid and based on factors not captured in the registry such as high bleeding risk, advanced disease or frailty. Treatment decisions might also vary based on lesion anatomy, lesion location and/or functional assessment. Despite reflecting real-world practice, the absence of systematic invasive physiology supporting decision of PCI is a limitation. Unfortunately, data on symptom improvement are not available in the SCAAR registry but would have provided important insight. We used the Swedish Prescribed Drugs Registry to access data on guideline-recommended MT and the prescription of a drug does not ensure patient adherence/compliance. We collected data on prescribed MTs 180 days prior to 7 days after angiography as we aimed to only include patients on guideline-derived MT on index date. However, the PCI group comprised both ad hoc PCI and patients undergoing elective PCI within 45 days. For patients undergoing elective PCI, an increase in prescribed MTs is believable, particularly for P2Y12i/clopidogrel. The SCAAR registry does not capture periprocedural MI, assessing periprocedural MI would benefit the MT group and as such possibly alter the results.

Conclusion

In this study, which sought to evaluate the outcomes of patients with CCS using a hierarchical approach, patients selected for revascularisation with PCI have improved outcome in terms of NACE and MACE, driven by lower rates of urgent revascularisation and MI. In addition, the use of PCI was associated with lower rates of cardiovascular mortality.

Data availability statement

No data are available. We can not make data available, but interested researchers can contact us and we may help with presenting aggregated data.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study was approved by the Regional Ethical Review Board in Lund (approval number: 2015/297) and research was carried out in accordance with appropriate ethical guidelines.

Acknowledgments

The authors thank the staff members of all PCI laboratories in Sweden for their continuous work of collecting data for the SCAAR registry.