Long-term outcomes of ischaemia with no obstructive coronary artery disease (INOCA): a systematic review and meta-analysis

Rationale

Chest pain is one of the most common presenting complaints in outpatient visits and to the emergency department.1 2 Of patients who are referred for invasive coronary angiography, with or without ischaemia on non-invasive stress testing, 30–50% are found to have ischaemia and no obstructive coronary artery disease (INOCA).3–5 Despite the confirmed absence of epicardial coronary artery disease (CAD), many patients experience recurrent presentations for the evaluation of chest pain.2 5 Physiological assessment of patients with INOCA has revealed endotypes, which include epicardial or microvascular vasospasm, coronary microvascular dysfunction or diffuse epicardial atherosclerosis in more than 75% of patients.4

While historically considered a benign syndrome, recent studies evaluating the long-term outcomes of patients with INOCA have reported a nearly fourfold increase in the incidence of mortality and major adverse cardiovascular events (MACEs) compared with patients without a diagnosis of INOCA.6–8 Previous studies evaluating INOCA have included a heterogeneous population of patients with chest pain and no obstructive CAD, patients with a specific diagnosis based on non-invasive testing and patients with a diagnosis based on invasive functional testing.9–14 Therefore, there are gaps in our understanding of INOCA and uncertainty regarding differences in the incidence of adverse events based on the underlying diagnosis (vasospastic angina (VSA) or microvascular angina (MVA)) or method of diagnosis (invasive or non-invasive).

Objectives

In this meta-analysis, we sought to determine the long-term prognosis of patients with INOCA stratified by both the underlying diagnosis and the method of diagnosis.

Search strategy

We conducted a systematic review and meta-analysis of published studies using a predefined protocol registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42021275070).

We searched Medline, Embase, Web of Science and Cochrane databases using a Boolean search strategy consisting of both keywords and controlled vocabulary combining terms for INOCA including ‘microvascular disease’, ‘coronary vasospasm’, ‘no obstructive coronary artery disease’, ‘microcirculatory’ and ‘coronary’ with ‘outcomes’, ‘major adverse events’ and ‘mortality’. The search was restricted to studies published in English language from January 1984 to August 2023. The complete search strategy is provided in online supplemental appendix table 1.

Study selection criteria

Studies were selected if they reported prespecified outcomes in patients with INOCA. Only studies that reported non-obstructive CAD (<70% narrowing) using anatomical evaluation (coronary CT angiography (CTA) or invasive angiography) were selected. Additionally, studies had to meet the inclusion criteria based on the Coronary Vasomotor Disorders International Study Group (COVADIS) guidelines (online supplemental appendix tables 2, 3).15 16

Studies were excluded if (1) more than 10% of study participants had CAD or prior or current acute coronary syndrome (ACS) presentations; (2) patients were diagnosed with less defined coronary abnormalities, such as diffuse (but non-obstructive) atherosclerosis, endothelial dysfunction, elevated resting flow and myocardial bridging; (3) the study included uncharacterised endotypes such as ‘angiographic slow flow’ that do not meet contemporary criteria for a disorder of coronary vasomotion and non-coronary surrogate testing (such as brachial artery flow mediated vasodilation, cold pressor test, handgrip test, etc).

Two investigators (NO and SSK) independently screened titles and abstracts generated by database search for eligibility criteria (figure 1). Additionally, two investigators (NO and AS) screened review records and selected relevant review articles whose bibliographies were screened for potentially relevant studies. If a study was considered potentially relevant, a full-text article was reviewed by NO and AS.

Request permission

Data abstraction

Two investigators (NO and AS) independently extracted the data into a Microsoft Excel spreadsheet and crosschecked for accuracy. The following fields were collected: first author first name and last name, year of publication, number of subjects, years of enrolment, study population baseline demographics including gender and cardiovascular risk factors, modality used to rule out obstructive CAD and method and modality of diagnosing specific INOCA subtype (invasive or non-invasive, specific type of test). Outcome measures were extracted with annual event rates for up to 5 years of follow-up.

Prespecified outcomes

We investigated two composite outcomes: (1) all cause-death and myocardial infarction (MI); (2) MACEs, defined as a composite of cardiovascular death, MI, stroke, hospitalisation for cardiovascular causes or coronary revascularisation.

Statistical analysis

Statistical analysis was performed by ZZ and HP. The results for each outcome were presented as incidence rates with 95% CIs.

Incidence rates from eligible studies were pooled using the DerSimonian-Laird random effects model. Freeman-Tukey double arcsine transformation was used to handle low incidence rates and to stabilise the variance. Degree of heterogeneity was estimated using I² statistics with the criteria: none ‘<25%’, low ‘25–49%’, moderate ‘50–74%’, and high ‘75+%’. Publication bias and small study effect were assessed by degree of funnel plot asymmetry and Egger’s regression test. The estimated missing studies were imputed by trim-and-fill method. Sensitivity analysis was conducted by sequentially excluding individual studies and the influence of each study on the overall incidence rate was examined. To examine sources of between-study variability, we performed subgroup analysis and random-effects meta-regression. All statistical analyses were performed using Stata V.17.0 (Stata Corp, College Station, Texas, USA).

Vasospastic angina

Twenty-one studies (7523 patients) were included for VSA (online supplemental appendix table 4). All studies ruled out obstructive CAD with coronary angiography. One study used the clinical diagnosis of VSA, 16 studies established the diagnosis of VSA invasively and 4 studies used either clinical or invasive criteria for subject inclusion. Among studies using invasive criteria, 6 used acetylcholine only, 12 used ergonovine or methylergonovine only, and 2 used either. Six studies used a cut-off ≥90% lumen reduction by visual assessments to define spasm and the remaining 15 studies used a cut-off that was lower than that or multiple cutoffs. Definitive VSA based on the COVADIS criteria was diagnosed in 5 studies and possible in the remaining 16 studies. The combined incidence of all-cause death and MI was 0.7 (95% CI 0.4 to 1.0)/100 patient-years (I²=90.3%, p<0.01) (figure 3) and a combined incidence of MACE was 1.1 (95% CI 0.5 to 1.9)/100 patient-years (I²=97.3%, p<0.01) (figure 4). There was no difference in either of the two composite outcomes when we compared studies defining spasm as ≥90% lumen reduction versus studies using lower a cut-off or when we compared studies with definitive versus possible VSA by the COVADIS criteria (p>0.05) (online supplemental appendix tables 7, 8).

Request permission

Request permission

Coronary microvascular angina

Thirty-nine studies (9779 patients) met the inclusion criteria for coronary MVA (online supplemental appendix table 5). One study used CTA, two studies used either CTA or angiography, and the remaining studies relied exclusively on coronary angiography to rule out obstructive CAD. The diagnosis of suspected MVA was present based on positive stress tests in 20 studies, while 17 studies had evidence of impaired coronary microvascular function: 11 had impaired coronary flow reserve (CFR), 5 had microvascular spasm and 1 study had an abnormal index of microcirculatory resistance. Two studies included participants meeting both criteria for suspected and definitive MVA but did not provide outcomes separately for the groups. The combined incidence of all-cause death and MI was 1.0 (95% CI 0.6 to 1.4)/100 patient-years (I²=92.2%, p<0.01) (figure 3) and a combined incidence of MACE was 2.5 (95% CI 1.6 to 3.6)/100 patient-years (I²=96.7%, p<0.01) (figure 4).

Patients whose diagnosis was established based on a positive stress test of any type had 5.7 times lower rates of all-cause death and MI (0.5 (95% CI 0.1 to 1.1) and 1.1 (95% CI 0.5 to 2.0) for exercise and imaging, respectively) than patients who had an abnormal CFR (4.7 (95% CI 2.0 to 8.4), p<0.01). When the outcomes were expanded to MACE, this trend persisted, but without statistical significance. The incidence of MACE was higher in the MVA cohort compared with the incidence in patients with VSA (2.5 (95% CI 1.6 to 3.6) vs 1.1 (95% CI 0.5 to 1.9), p=0.025) (online supplemental appendix tables 7, 8).

Publication bias

We observed no funnel plot asymmetry for all-cause death and MI (online supplemental appendix figure 1, Egger’s test: p=0.154), but the presence of funnel plot asymmetry for MACE (online supplemental appendix figure 2, Egger’s test: p=0.022) indicated the presence of publication bias. A sensitivity analysis was performed with the leave-one-out method. There were no studies that substantially influenced the overall incidence estimate (online supplemental appendix figures 3, 4).

Additional analyses

We included continuous variables hypertension, hyperlipidaemia and diabetes in the meta-regression to further characterise between-study variance. On average, patients with hypertension, hyperlipidaemia or diabetes had higher MACE rates (p<0.01). For the incidence of all-cause death and MI, the between-study variance can be explained by hypertension (20.2%), hyperlipidaemia (16.1%) and diabetes (28.2%); for the incidence of MACE, the between-study variance can be explained by hypertension (38.6%) and diabetes (18.4%) via the R² statistics. However, high residual variation due to between-study heterogeneity remained (I²>90%), which may be due to other covariates or uncharacterised factors (online supplemental appendix tables 9,10).

Limitations

Our meta-analysis has limitations including high heterogeneity among the included studies, many of which were single centre studies. The overall quality of data is poor and few studies fulfil the criteria for ‘definitive’, while most can only be classified as ‘possible’ VSA or MVA. Additionally, it is likely that an unknown burden of atherosclerotic coronary artery disease beyond angiographic assessment existed; therefore, the contribution of coronary artery disease to the reported associations remains incompletely characterised.