Objectives Utility of CT coronary angiography (CTA) and coronary artery calcium (CAC) scoring in risk stratification prior to non-cardiac surgery is unclear. Although current guidelines recommend stress testing in intermediate-high risk individuals, over one-third of perioperative major adverse cardiovascular events (MACE) occur in patients with a negative study. This systematic review and meta-analysis evaluates the value of CTA and CAC score in preoperative risk prognostication prior to non-cardiac surgery.
Methods MEDLINE, PubMed and EMBASE databases were searched for articles published up to June 2018. Summary ORs for degree of coronary artery disease (CAD) and perioperative MACE were pooled using a random-effects model.
Results Eleven studies were included. Two hundred and fifty-two (7.2%) MACE occurred in 3480 patients. Risk of perioperative MACE rose with the severity and extent of CAD on CTA (no CAD 2.0%; non-obstructive 4.1%; obstructive single-vessel 7.1%; obstructive multivessel 23.1%, p<0.001). Multivessel disease (MVD) demonstrated the greatest risk (OR 8.9, 95% CI 5.1 to 15.3, p<0.001). Increasing CAC score was associated with higher perioperative MACE (CAC score: ≥100 OR 5.1, ≥1000 OR 10.4, both p<0.01). In a cohort deemed high risk by established clinical indices, absence of MVD on CTA demonstrated a negative predictive value of 96% (95% CI 92.8 to 98.4) for predicting freedom from MACE.
Conclusions Severity and extent of CAD on CTA conferred incremental risk for perioperative MACE in patients undergoing non-cardiac surgery. The ‘rule-out’ capability of CTA is comparable to other non-invasive imaging modalities and offers a viable alternative for risk stratification of patients undergoing non-cardiac surgery.
Trial registration number CRD42018100883
- coronary angiography
- non-cardiac surgery
- postoperative complications
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.
Cardiovascular complications are among the leading causes of morbidity and mortality in patients undergoing non-cardiac surgery.1 Despite existing preoperative assessment strategies, our capacity to predict these events in individual patients is limited.2–4 Current guidelines recommend stress testing for patients undergoing intermediate-to-high risk non-cardiac surgery with an unknown or impaired functional status.2 5 While this approach facilitates detection of obstructive coronary artery disease (CAD) causing myocardial ischaemia, it fails to identify subclinical atherosclerosis that may be a substrate for perioperative plaque rupture and thrombosis.6
CT coronary angiography (CTA) is a non-invasive modality that can quantify atherosclerotic plaque burden and stenosis, while coronary artery calcium (CAC) score is an established marker of atherosclerotic disease burden. The diagnostic accuracy and prognostic ability of CTA and CAC is well described.7–9 However, there remains clinical equipoise on its role in preoperative risk stratification with current guidelines citing insufficient evidence to make definitive recommendations.2 Some studies have demonstrated that CTA and CAC score may predict the occurrence of perioperative major adverse cardiovascular events (MACE)10–13 However, the overall strength of association and the degree of CAD that confers a high risk in this setting is unclear. Furthermore, evidence from individual studies is limited by the uncertainty surrounding risk estimates, both as a result of a limited number of outcomes and the differences in the patient populations evaluated. Therefore, we conducted a systematic review and meta-analysis to assess the ability of preoperative CTA and CAC score for cardiovascular risk stratification in patients undergoing non-cardiac surgery.
A literature search was performed through the MEDLINE, EMBASE and PubMed databases up to June 2018. Keywords using Medical Subject Heading (MeSH), where available, included coronary artery disease, atherosclerosis, multidetector computed tomography, coronary CT angiography, coronary calcium score, non-cardiac surgery, preoperative, perioperative, prognosis, mortality and major adverse cardiovascular events. The search was not limited by language nor date of publication. MeSH terms used for the MEDLINE search strategy are presented in online supplementary table S1. Reference lists of reviewed articles were screened to identify further relevant studies. The study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement14 and was prospectively registered with the PROSPERO international register.
Study inclusion and exclusion criteria
Inclusion criteria for studies were as follows: (1) patients undergoing intermediate-to-high risk non-cardiac surgery, (2) preoperative CTA or CAC score performed for assessment of CAD, (3) evaluation of the association between extent and severity of CAD/CAC and perioperative MACE and (4) postoperative clinical follow-up of at least 30 days. We excluded studies performed in patients undergoing any cardiac surgery and studies evaluating patients with cardiac symptoms or those with acute coronary syndromes.
We prespecified variables for retrieval prior to the literature search. Two investigators (ANK and FJH) independently searched and extracted relevant articles which was subsequently verified by the senior investigator (OF) with discrepancies resolved by consensus. The following variables were extracted: study design and characteristics, follow-up duration, inclusion and exclusion criteria, sample size, operation details, type of CT scanner, age, known or suspected CAD and percentage of male gender. Preoperative clinical risk classification using the Revised Cardiac Risk Index (RCRI) was collated when reported. This relies on the presence or absence of the following six variables: high-risk surgery (intrathoracic, intraperitoneal or suprainguinal vascular surgery), ischaemic heart disease, congestive heart failure, insulin therapy with diabetes mellitus, cerebrovascular disease and renal dysfunction (preoperative serum creatinine >2.0 mg/dL).15 RCRI ≥3 was considered high risk.15
Degree of coronary artery stenosis was categorised into: no CAD (absence of atherosclerosis), non-obstructive (<50% stenosis) and obstructive CAD (≥50% stenosis) as reported in the studies. Multivessel disease (MVD) was defined as ≥2 vessels with obstructive CAD. Coronary calcium score reporting varied across studies due to differences in categorical classification and categorised as 0, 1–399, ≥400 and ≥1000. Studies with no perioperative MACE were excluded from the quantitative meta-analysis.16
Clinical outcome assessment
Perioperative MACE was categorised according to the definitions from each study. This included both individual and composite outcomes of cardiac death, non-fatal myocardial infarction (MI), congestive cardiac failure or haemodynamically significant ventricular arrhythmia or complete heart block. Diagnosis of MI in most studies required a troponin rise with one of the following: ischaemic signs or symptoms, ischaemic changes on electrocardiography or a new imaging abnormality suggestive of an MI. Definitions of cardiac death included sudden death as well as death secondary to MI, heart failure or fatal arrhythmias. Perioperative MACE at 30 days was extracted from each study, when reported.
Methodological quality was assessed by two independent investigators using the Newcastle-Ottawa Scale (NOS) for quality assessment of observational studies in meta-analyses.17 Assessment of quality on this instrument was judged on study group selection, study group comparability and outcome assessment. Studies meeting ≥5 criteria were considered to be high quality.17
Descriptive statistics are presented as mean and SD for continuous variables and absolute and relative frequencies for categorical variables. To determine the pooled risk associated with severity of CAD (extent of stenosis and number of vessels involved) as well as CAC score, summary estimates for ORs with 95% CIs were calculated using the Der Simonian and Laird random-effects model.18 Statistical heterogeneity was quantified using the I2 statistic, where I2≥80% was considered to be significant interstudy heterogeneity.19 We collected the number of events in each study to calculate summary sensitivity, specificity, hierarchical summary receiver-operating characteristic (HSROC) curve and likelihood ratios. We utilised a Fagan’s nomogram to graphically depict how the results of a CTA would affect the probability of perioperative MACE. This is a Bayesian graphical tool that estimates how the result of a diagnostic test changes the probability of a patient having a condition.20 A line drawn from the pretest probability through the likelihood ratio of interest intercepts the new posttest probability for the patient. A pretest probability of 11% was chosen as it corresponds to an RCRI of 3 based on prior publications.15 To compare the incidence of perioperative MACE based on severity of CAD on CTA, χ² test of independence was performed with Bonferroni post hoc correction. As the number of included studies was <10 for each CTA and CAC analysis, we refrained from any tests on publication bias.21 We considered a p<0.05 as statistically significant. Statistical analyses was performed using Stata V.15/MP.
The initial search yielded 1882 studies. We eliminated 1814 after initial screening. Details of the literature search and excluded studies are reported in figure 1. Eleven studies with a total of 3480 patients were included in the final qualitative analysis.10–13 22–28 This included seven studies that evaluated CTA and seven studies of CAC score inclusive of three studies that evaluated both. Individual study characteristics and details about the types of surgery included are summarised in table 1 and online supplementary table S2. Three CTA studies and two CAC studies were excluded from the meta-analysis. CTA studies were excluded due to study eligibility requisite on a negative stress test,22 no perioperative MACE occurrence28 and no quantitative data available for meta-analysis.26 The CAC studies were excluded from quantitative analysis due to a MACE definition that included any troponin rise even in the absence of a clinical MI27 and another where no perioperative MACE occurred.28
Mean age of study participants was 65±8 years and 67% were male. Baseline characteristics with RCRI scores are presented in online supplementary table S3 and individual study definitions of MI are reported in online supplementary table S4. On assessment of study quality based on the NOS (see online supplementary table S5), five of the seven studies of CTA were of high quality (median score 6, range 3–8). Studies of CAC were generally of lower quality (median score 5, range 4–7).
A total of 252 (7.2%) perioperative MACE outcomes were reported (table 2). This included 71 cases of cardiac death, 142 cases of non-fatal MI, 16 cases of clinically significant heart failure and 23 cases of arrhythmia. One study included unplanned urgent revascularisation following surgery and ischaemic stroke as perioperative MACE.25 Three studies10 24 27 routinely measured troponin levels for 3 days postoperatively. Diagnosis of MI was only confirmed with concurrent clinical symptoms or ischaemic ECG changes in two of these studies.10 24 The outcome assessment was blinded in two studies10 24 with the remainder being assessed by clinician adjudication.
Coronary stenosis and perioperative MACE
Seven studies reported perioperative outcomes and associated preoperative CTA findings. Occurrence of perioperative MACE rose incrementally with the severity of CAD as stratified on CTA (figure 2). Four of the seven studies were included in the quantitative meta-analysis with study exclusion detailed previously. Compared with absence of CAD, multivessel obstructive disease on CTA demonstrated the highest risk of perioperative MACE (OR 8.9, 95% CI 5.1 to 15.3, p<0.001, I2=0%). Significant, although less risk was also demonstrated for single-vessel obstructive disease (OR 3.0, 95% CI 1.5 to 5.8, p=0.002, I2 =0%). There was no significant association between non-obstructive CAD and MACE (OR 1.7, 95% CI 0.8 to 3.7, p=0.16, I2=0%). Forest plots illustrating these findings are presented in figure 3.
CAC score and MACE
Five studies reported CAC score values with associated risk of perioperative MACE and were included in the quantitative analysis. Due to variability in CAC cut-offs between studies, adjusted threshold CAC values and corresponding MACE outcomes were extracted. On pooled analysis, increasing CAC demonstrated higher perioperative risk with a calcium score ≥1000 conferring the highest risk (OR 10.4, 95% CI 1.6 to 69.7, p<0.01) (table 3).
Predictive value of CTA in addition to clinical risk indices
The utility of CTA as an adjunct to clinical risk classification by the RCRI was evaluated using data from three studies that concurrently reported RCRI scores, severity/extent of CAD and associated MACE.11 12 23 In a high-risk cohort (RCRI ≥3) dichotomised into MVD and no MVD, CTA demonstrated a pooled negative predictive value (NPV) of 96% (95% CI 92.8 to 98.4, I2=15%) with low heterogeneity. Similar pooled estimates in obstructive CAD alone could not be calculated due to differences in study outcome reporting. Pooled likelihood ratios were unable to be estimated in the obstructive CAD subgroup due to low study numbers.
Diagnostic performance of presence or absence of CAD
Test parameters assessed the presence or absence of any CAD as well as obstructive CAD, for predicting perioperative MACE. Analysis of pooled performance measures for the presence of any CAD demonstrated a sensitivity of 0.89 (95% CI 0.75 to 0.95) and specificity 0.35 (95% CI 0.16 to 0.61) (figure 4A), positive likelihood ratio (+LR) of 1.4 (95% CI 1.03 to 1.82) and negative likelihood ratio (−LR) of 0.3 (95% CI 0.20 to 0.52). Low specificity was reported particularly in one study that likely accounted for the high heterogeneity.10 In patients with obstructive CAD compared with non-obstructive/normal coronaries, test parameters demonstrated higher specificity and positive likelihood ratio with lower sensitivity (sensitivity of 0.7 [0.53–0.85], specificity of 0.6 [0.42–0.77], +LR 1.9 [95% CI 1.42 to 2.65] and –LR 0.47 [95% CI 0.32 to 0.70]).
The HSROC analysis for CAD on CTA to predict perioperative MACE demonstrated an area under the curve of 0.77 (95% CI 0.73 to 0.80) (figure 4B). Figure 5 depicts how a –LR and +LR affect posttest probability after CTA. In a hypothetical patient with an RCRI score of 3 that confers a perioperative risk of ~11%, no CAD on CTA would result in a posttest probability of 4% (figure 5). While the posttest probability in those with any CAD is modest due to the low +LR, the risk of perioperative MACE would increase to 33% in a patient with multivessel obstructive CAD (see online supplementary figure S1).
Our findings demonstrate that the severity and extent of CAD poses an incremental risk for perioperative MACE with an eightfold increase noted in patients with obstructive MVD. Non-obstructive or no CAD was found to convey a modest risk a modest risk of perioperative MACE in patients with a high RCRI score. To our knowledge, this is the first systematic review and meta-analysis to evaluate the ability of CTA and CAC for the prediction of perioperative cardiovascular events in non-cardiac surgery. These results suggest a potential role for CTA in informing clinicians about cardiovascular risk stratification prior to non-cardiac surgery.
The utility of CTA and CAC score as a perioperative risk stratification tool is unclear. This is reflected in current guidelines that highlight the paucity of evidence in this area.5 Although abnormalities on stress testing may confer high perioperative risk, over one-third of perioperative MACE occur in patients with a negative study.4 CAC scoring and CTA are robust tools for the quantification of atherosclerotic plaque and the severity of CAD without physiological or pharmacological stress. This may be advantageous in a preoperative cohort that may be unable to exercise such as those undergoing orthopaedic or vascular surgery, or those with contraindications to pharmacological stress testing agents.12
Our findings demonstrate that obstructive CAD, particularly in multiple coronary arteries, conferred a significant risk for perioperative MACE. Multivessel coronary disease identifies a cohort with a higher overall burden of atherosclerosis. These patients are thus at risk of both demand–supply mismatch ischaemia and plaque rupture that may be potentiated in a perioperative setting characterised by neurohormonal dysregulation, inflammation and coronary shear stress.6 On pooled analysis, 23% of patients with multivessel CAD experienced perioperative MACE. Shalaeva et al 23 report that this hazard exceeded 50% in a cohort of patients with diabetes that underwent vascular surgery. As such, multivessel CAD detected by CTA may identify a group that is at high risk for undergoing elective surgery. Future research that assess vulnerable plaque characteristics or CT perfusion in addition to luminal stenosis may further augment the value of CTA in perioperative risk stratification.
The absence of CAD on CTA conveys a low-modest risk of perioperative MACE (–LR of 0.3 (95% CI 0.2 to 0.5) in our study. This is consistent with studies that have used stress echocardiography (–LR 0.2, 95% CI 0.2 to 0.3) and stress myocardial perfusion imaging (−LR 0.4, 95% CI 0.4 to 0.5) in risk stratification prior to non-cardiac surgery.4 The clinical utility of this finding, however, may be limited as most patients at intermediate-high clinical risk often have a degree of subclinical atherosclerosis. Although the risk conferred by non-obstructive CAD alone did not meet statistical significance on pooled analysis, combining both non-obstructive and no CAD as a ‘negative study’ adversely affected CT’s ‘rule-out’ capability with only a modest −LR (0.47, 95% CI 0.32 to 0.70) for predicting freedom from MACE. This may reflect the underlying mechanisms of a perioperative MI caused by disruption of a non-obstructive lipid-rich plaque.29 Alternatively, it may demonstrate the heterogeneous pathophysiology of perioperative MACE, where factors extraneous to flow-limiting CAD such as anaemia, sepsis and hypovolaemia, also trigger cardiovascular events.6
Although adverse events were rare among patients without significant CAD, the majority of patients with coronary disease did not experience MACE in the 30-day period following surgery. Indeed, results from the multicentre VISION CT-study by Sheth et al 10 reported on inappropriate overestimation of risk, based on preoperative CTA findings. However, the low event rates and estimated predictive value of CTA in this study may be reflective of the cohort assessed, where >75% of patients had an RCRI score ≤1. Based on current guidelines,5 these patients could have undergone surgery without any preoperative cardiovascular investigations. Clinical risk prediction tools in this context should best be used as ‘gatekeepers’ of downstream non-invasive testing, where only those with intermediate-to-high RCRI scores undergo further investigation.30 Data pooled from three studies in our analysis demonstrated a NPV of 96% in an RCRI ≥3 cohort in the absence of multivessel CAD. Utilising CTA in this high-risk cohort can reduce overestimation of risk based on clinical indices alone. Furthermore, this can reduce patients opting out of undergoing surgery due to a higher perceived risk and ensure that heightened postoperative surveillance strategies such as troponin measurements and cardiac monitoring are reserved for patients whoare at highest risk.31
We also evaluated the role of CAC score in preoperative risk stratification. There was variability in the CAC score cut-offs between studies that likely contributed to the somewhat non-linear relationship of CAC score with perioperative cardiovascular events. Despite this, the greatest risk of MACE was observed in those with a CAC score ≥1000 and likely represents individuals with a higher burden of atherosclerosis. CAC score may also be considered to avoid CTA in cases of patients with a CAC score of 0 or a very high CAC. One study that assessed comparative additive value of CTA and CAC did not demonstrate a significant difference with the addition of CTA to CAC findings (p=0.71), although the low event rates suggest that it may have been underpowered to assess for this difference.11 A CAC score likely represents a continuum of risk in the perioperative setting. Whether addition of CTA findings to CAC score offers additional risk predictive capability in patients who undergo non-cardiac surgery remains unresolved.
While this study describes an association between severity of CAD and perioperative MACE, the exact translation of our study findings to clinical practice is unclear. No CAD on CTA likely confers a favourable perioperative prognosis, while multivessel CAD identifies a high-risk cohort. Whether findings of non-obstructive or single-vessel disease on CTA in a majority of screened patients changes clinical practice remains uncertain. Prior studies have not demonstrated a benefit of either preoperative revascularisation or medical therapy in mitigating MACE.32–34 However, the recent findings of the Management of Myocardial Injury After Noncardiac Surgery Trial (MANAGE) trial has shown that dabigatran lowered the risk of major vascular complications in patients with myocardial injury after non-cardiac surgery, suggesting that medical therapy can alter outcomes.35 In a non-surgical setting, an analysis of the Scottish Computed Tomography of the Heart Trial (SCOT-HEART) trial showed that CTA-guided institution of preventative aspirin and statin therapy demonstrated up to 50% reduction in MI.36 37 It is conceivable that CTA in the preoperative setting could be used to guide initiation of preventive drug therapies, but future studies to assess this concept are imperative. Despite the utility of CTA, there remains concerns with the potential costs of this test. Although cost-effectiveness analysis of CTA has yielded comparable results to a stress testing strategy in patients evaluated for chest pain, this has not been performed in an asymptomatic cohort undergoing non-cardiac surgery.38 Further, as studies have not compared CTA to stress testing in non-cardiac surgery, it is unclear whether one modality may be superior in risk stratification.
We acknowledge certain limitations in this study. First, the shortcomings of a meta-analysis may be even more pronounced when evaluating diagnostic test accuracy due to the variety of definitions in clinical end-points in a heterogenous patient population.16 We have sought to control for this by using standardised perioperative end-points of cardiac death, MI, heart failure and haemodynamically significant arrhythmias. Second, only four CTA studies were included in the quantitative meta-analysis. Despite this shortcoming, the studies were of good quality which would improve generalisability of our findings. Third, physicians were blinded to the results in only two studies. Following on, it remains unclear how many patients with an abnormal result underwent invasive coronary angiography and revascularisation, or the number of patients with high-risk coronary anatomy who did not undergo surgery altogether. Fourth, different non-cardiac operations are associated with varied risk of MACE.2 Due to low individual study events, perioperative MACE was not reported and stratified by type of surgery. Furthermore, due to differences in study outcome reporting, we were unable to compare the incremental risk conferred by CTA over the RCRI score. Despite significant heterogeneity in estimation of sensitivity and specificity, sensitivity analysis by individual study exclusion could not be performed due to low study numbers.
Finally, routine postoperative troponin levels were measured in only three studies. As a majority of perioperative MIs occur in the absence of symptoms,3 these results may underestimate the true event rates and the risk conferred by abnormal CTA findings.
Severity and extent of CAD on CTA conferred incremental risk for perioperative MACE in patients undergoing non-cardiac surgery. The rule-out capability of CTA in intermediate-to-high risk patients is comparable to other non-invasive imaging modalities and may be a viable alternative in perioperative cardiovascular risk stratification.
What is already known on this subject?
Atherosclerosis on CT coronary angiography (CTA) and coronary calcium scoring (CAC) confer adverse prognosis in patients evaluated for chest pain.
However, current guidelines cite insufficient evidence supporting their use in risk stratification prior to non-cardiac surgery.
What might this study add?
When compared with no coronary artery disease, patients with single-vessel and multivessel obstructive coronary disease on CTA demonstrated a threefold and eightfold increased risk of perioperative major adverse cardiovascular events (MACE), respectively.
Absence of multivessel disease on CTA yielded a negative predictive value of 96% in patients deemed high risk by clinical indices, comparable to other non-invasive modalities.
Increasing CAC score also conferred increased risk for perioperative MACE.
How might this impact on clinical practice?
Severity and extent of coronary disease on CTA may identify patients at a high risk for cardiac complications after elective non-cardiac surgery.
It also demonstrates a ‘rule out’ capability comparable to stress testing modalities.
CTA may offer a viable alternative in risk stratifying patients undergoing non-cardiac surgery, but further prospective evaluation is required to establish its role.
Contributors ANK and FJH conducted data acquisition and analysis and drafted the manuscript. H-CH, AWT, HSL, FMA-I, PJG and OF supervised study design, data interpretation and assisted with drafting and reviewing of manuscript.
Funding ANK is a recipient of the National Health and Medical Research Council of Australia/National Heart Foundation Post-Graduate Scholarship and Royal Australasian College of Physicians Blackburn Scholarship. AWT is a recipient of the Early Career Fellowship from the National Health and Medical Research Council of Australia. HSL is supported by the Neil Hamilton Fairley Early Career Fellowship from the National Health and Medical Research Council of Australia.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Not required.