CT coronary angiography guided treatment versus routine invasive coronary angiography in patients with non-ST elevation myocardial infarction: a systematic review and meta-analysis

Abstract

Background

Non-ST-segment-elevation myocardial infarction (NSTEMI) accounts for the majority of acute coronary syndrome presentations. Invasive coronary angiography (ICA) is recommended but challenging to deliver within guideline-recommended timeframes and unnecessary in patients without obstructive disease. CT coronary angiography (CTCA) offers a non-invasive alternative for lower-risk patients that may reduce procedural-related complications and optimise resource use by avoiding unnecessary ICA.

Methods

We conducted a systematic review and meta-analysis following Cochrane standards. Four databases were searched from January 2005 to March 2025. Eligible studies included adult patients with NSTEMI undergoing CTCA to guide management, with comparators of standard care (ICA-first strategies). Primary outcomes were ICA utilisation, composite cardiovascular events (myocardial infarction, cardiovascular and all-cause death) and diagnostic accuracy of CTCA versus ICA.

Results

From 12 058 records, eight studies met inclusion criteria; three randomised controlled trial (RCT) studies and five observational cohorts; total n≈2700. Across two RCTs, a CTCA-first strategy did not significantly reduce ICA use (pooled relative risk (RR)=0.82 (0.55–1.22)). Evidence for composite clinical outcomes was heterogeneous: The Rapid Assessment of Potential Ischaemic Heart Disease with CTCA trial reported no difference in death or MI, whereas the trial, Computed Tomography Cardiac Angiography Before Invasive Coronary Angiography in Patients with Previous Bypass Surgery, observed reduced events at 1 year (RR=0.63 (0.43–0.94)). Pooled diagnostic accuracy for >50% stenosis from observational studies demonstrated high sensitivity (96.3%) and moderate specificity (79.6%), with high heterogeneity. Length of stay and patient satisfaction were similar or improved with CTCA-first strategies, though cost-effectiveness data were limited. Overall evidence quality ranged from moderate (RCTs) to low (observational studies).

Conclusions

CTCA demonstrates excellent diagnostic accuracy and may reduce complications while maintaining clinical outcomes in patients with NSTEMI. Current evidence does not show consistent reductions in ICA use or major adverse events. Larger trials are needed to clarify its role in NSTEMI management pathways.

PROSPERO registration number

CRD420251003903

WHAT IS ALREADY KNOWN ON THIS TOPIC

• CT coronary angiography (CTCA) has diagnostic accuracy comparable with invasive coronary angiography (ICA) for detecting obstructive coronary artery disease and improves diagnosis, treatment targeting and outcomes in patients with stable chest pain. There is growing interest in whether CTCA could deliver similar benefits and be a gatekeeper for ICA in patients with non-ST elevation myocardial infarction (NSTEMI).

WHAT THIS STUDY ADDS

• This systematic review and meta-analysis examined diagnostic accuracy and clinical outcomes of a CTCA-first approach in patients with NSTEMI. CTCA demonstrated excellent diagnostic accuracy in this setting and reduced the use of ICA in the only randomised trial to use it as dedicated upfront strategy prior to ICA. Despite good evidence for diagnostic accuracy and safety, the systematic review highlights the high degree of heterogeneity between existing studies, limiting interpretation of any impact on clinical outcomes, ICA use and patient satisfaction.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The available evidence indicates that CTCA is safe and has potential utility in selecting patients with NSTEMI for ICA. However, owing to limited data and substantial heterogeneity among existing trials, larger-scale clinical studies are required to determine whether a CTCA-first strategy confers benefits in clinical outcomes, patient satisfaction and resource utilisation relative to the current standard of care.

Introduction

Non-ST-segment elevation myocardial infarction (NSTEMI) represents the majority of acute coronary syndrome presentations, with considerable morbidity, mortality and healthcare burden worldwide.¹ Accurate risk stratification and timely decision-making are required to optimise outcomes in this heterogeneous patient population.² Patients with ongoing ischaemic chest pain and dynamic ECG changes are at highest risk and should have immediate invasive coronary angiography (ICA).³ The implementation of high-sensitivity cardiac troponin testing has increased NSTEMI diagnosis by 27% to include lower risk presentations that require further investigation.⁴ Current guidelines recommend ICA for patients with NSTEMI to assess coronary anatomy and guide revascularisation decisions. The European Society of Cardiology recommends ICA within 24 hours and UK guidelines within 72 hours of admission.^{3 5} These target time windows are difficult to meet for an enlarging patient population, particularly when admission is to hospitals without invasive cardiac facilities.⁶ In addition, ICA carries procedural risks, is resource-intensive and is unnecessary for patients with non-obstructive coronary artery disease.

CT coronary angiography (CTCA) has emerged as a non-invasive imaging modality capable of providing high-resolution visualisation of coronary anatomy. With its proven diagnostic accuracy in stable chest pain populations,⁷ upfront CTCA, prior to a decision to proceed with ICA, improves clinical outcomes by ensuring patients receive the correct treatment⁸ and reducing complications associated with routine ICA.⁹ There is growing evidence supporting its use in the setting of acute MI. CTCA has been proposed as a potential alternative or adjunct to ICA in selected patients with NSTEMI.^{10 11} By enabling early exclusion of obstructive coronary artery disease, CTCA may facilitate more efficient triage, reduce unnecessary invasive procedures and optimise resource utilisation.

Despite increasing interest and several studies evaluating CTCA in the context of NSTEMI, the role of this modality in guiding clinical management remains unclear. Variation in study designs, patient populations and outcome measures has limited the ability to draw definitive conclusions about its utility. A comprehensive synthesis of the existing evidence is therefore necessary to assess whether CTCA can safely and effectively inform management decisions in NSTEMI and impact clinical outcomes.

This systematic review and meta-analysis aims to evaluate the diagnostic performance, clinical impact and safety profile of CTCA in the management of patients with NSTEMI. Specifically, we assess its role in influencing downstream invasive procedures, hospital resource use and major cardiovascular outcomes compared with standard care pathways.

Methods

A systematic review was performed with standard protocol adopted from the Cochrane Handbook for Systematic Reviews.¹² Searches were limited to four databases. This review was registered on PROSPERO (CRD420251003903) on 17 March 2025.¹³ Six independent reviewers were involved from screening to data extraction.

Literature search

A systematic search on ‘OVID MEDLINE(R) All’, ‘Embase Classic+Embase’, ‘Cochrane Controlled Register of Trials (CENTRAL)’ and ‘clinicaltrials.gov’ was completed. Entry date limits were set from 1 January 2005 to 19 March 2025. Searches were performed with MeSH subject headers, EmTree terms and free text terms. In brief, these included terms such as “Acute Coronary Syndrome”, “Coronary Artery Disease”, “Non-ST Elevated Myocardial Infarction”, “Computed Tomography Angiography” and “Coronary Angiography”. Our search strategy is listed in online supplemental appendix table 1. Searches were performed with an English language restriction. The Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia) was used.¹⁴

Inclusion and exclusion criteria

The Population, Interventions, Comparator, Outcomes and Study Characteristics model¹⁵ was used to formulate the eligibility criteria. A pilot screen (n=7) was performed to confirm validity of criteria. The population was defined as patients >18 years presenting with a definitive diagnosis of NSTEMI. The intervention was the use of CTCA to guide management of NSTEMI. All studies involved follow-on ICA if clinically indicated. The comparator was standard care guided by initial invasive coronary angiogram. Risk stratification beyond diagnosis of NSTEMI was not performed. Patients with unstable angina or undifferentiated high-risk chest pain were not included.

Outcomes included between-group use of ICA, a combined clinical outcome of MI, all-cause death and cardiovascular death and diagnostic accuracy of CTCA versus ICA. Secondary outcomes included procedure-related complications, resource use including hospital length of stay, healthcare costs and patient satisfaction. Outcomes were obtained through direct data extraction. Studies were excluded if data were unavailable or there was insufficient clarity surrounding whether patients had a diagnosis of NSTEMI. If there were mixed cohorts of NSTEMI and patients with troponin-negative chest pain, studies were only included if the authors were successfully contacted to provide data on only NSTEMI patients. Full-text, English language papers presenting empirical data were included. The search was restricted to English language as the majority of high-quality evidence in the field is produced and disseminated in English. Data from secondary sources such as systematic reviews were not extracted but were used to identify papers that were not detected by the initial search strategy.

Title, abstract and full-text screen

Six reviewers independently reviewed titles, abstract and full-text articles selected with the above criteria. Each article was screened by two reviewers. Any conflicts were resolved by a third independent reviewer with discussion with the reviewers involved. Removal of duplicates was performed on Covidence.

Data extraction

A data extraction form was piloted. Data extraction for each article was conducted by two reviewers per article. A third reviewer checked for eligibility of data from articles marked for data extraction. Any conflicts were resolved by the third reviewer. Extracted data included study design, follow-up duration, whether the protocol allowed for discretionary use of ICA, feasibility to only extract data for patients with NSTEMI, population demographics, use of ICA during index admission after CTCA and clinical outcomes (all-case death, cardiovascular death, MI), treatment allocation after imaging (eg, medical management and percutaneous/surgical coronary revascularisation) and diagnostic accuracy (eg, true positive, true negative, false positive, false negatives). Additional data pertaining to patients diagnosed with NSTEMI were requested and obtained from the Rapid Assessment of Potential Ischaemic Heart Disease with CTCA (RAPID-CTCA)¹⁶ and Randomised Controlled Trial to Assess Whether Computed Tomography Cardiac Angiography Can Improve Invasive Angiography in Bypass Surgery (BYPASS-CTCA)¹⁷ investigators. Secondary outcomes included procedure-related complications, resource use (eg, length of hospital stay, healthcare costs) and patient satisfaction (eg, quality-of-life measures, chest pain symptoms, qualitative measures).

Risk-of-bias assessment

All studies were either observational cohort studies or randomised controlled trials (RCTs). We used the Cochrane Risk of Bias 2 (RoB 2) tool, Risk Of Bias In Non-randomised Studies - of Interventions (ROBINS-I) tool and Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 tool.18,20 Each article was assessed by two reviewers and discrepancies were discussed and adjudicated by a third reviewer.

Data synthesis

Given the limited availability of data for outcomes like ICA use and other clinical endpoints, we conducted a narrative synthesis of the extracted findings. Risk ratios were calculated or obtained from authors. Due to study population heterogeneity, RCTs were analysed individually. Analyses of risk ratios and diagnostic accuracy were carried out in R, employing an inverse-variance random-effects approach. Pooled estimates were derived using the DerSimonian-Laird method. Continuity corrections were performed if zero events were encountered.

A meta-analysis of diagnostic accuracy for a coronary diameter stenosis >50% was performed from observation studies. The criteria for inclusion for all observational studies were similar; pooled diagnostic accuracy and 95% CI data were calculated from data presenting true positives, true negatives, false positives and false negatives. Visual assessment of the forest plot and I² statistic was used to assess heterogeneity. Evidence quality was appraised with the Grading of Recommendation, Assessment, Development and Evaluation (GRADE) tool.²¹

Patient and public involvement

We have discussed this work with patients on the wards while creating a patient information video explaining a trial investigating a CTCA-first approach versus standard-of-care ICA.

Results

Summary of records

The search strategy (online supplemental appendix table 1) identified 12 058 records from four databases as shown in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram (figure 1).¹² The search was conducted on 19 March 2025 with English language filters. Databases included Ovid MEDLINE(R) All (n=5895), Embase Classic+Embase (n=5512), CENTRAL (n=590) and clinicaltrials.gov (n=61). Duplicate records (n=3188) were removed through Covidence and manual selection.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram¹² with the illustration of flow from initial search, title and abstract screening, full text screening and final count of records after each stage with reason for exclusion. RCT, randomised controlled trials.

From 8870 records, 8641 were excluded at the title and abstract screening stage. 229 records were included for full-text screening, from which three RCTs and five observational cohort studies were identified for data extraction (table 1). BYPASS-CTCA did not examine discretionary use of ICA following CTCA.¹⁷ Similarly, the five observational cohort studies did not report CTCA-guided use of ICA and focused on diagnostic accuracy.

Table 1. Details of the eight studies involving patients with NSTEMI selected for data extraction.

Study/country	Study name design/participant number/male sex (%)	Population/mean Age CTCA arm, years±SD	Population size CTCA, N (%)	Population size SoC, N (%)	Discretionary ICA	CTCA used to guide treatment	Reference (standard of care) versus index test	Outcome
Grey et al UK 22	RAPID-CTCA RCT n=1748 (n=689 with NSTEMI) 71% male	Patients presenting to the emergency department with acute chest pain and intermediate risk of ACS with one or more of previous coronary heart disease, raised levels of cardiac troponin or abnormal ECG. Mean age: 63±12 years	350 (51)	339 (49)	Yes	Yes	Routine care including ICA versus minimum 64-detector scanner	All cause death and myocardial infarction
Jones et al UK 17	BYPASS-CTCA RCT n=688 (n=310 with NSTEMI) 82% male	Patients with prior coronary bypass surgery referred for ICA ≥18 years of age. NSTEMI patients a subset of all included patients. Mean age: 70±11 years	155 (50)	155 (50)	No	Yes	Routine ICA versus third generation dual source scanner	Procedure time for ICA Patient satisfaction Contrast-induced nephropathy
Smulders et al Netherlands10	CARMENTA RCT n=207 62% male	Patients with NSTEMI referred for ICA. Absence of ST-T elevation and/or dynamic ST-T wave changes suggesting myocardial ischaemia and elevated hs-cTnT levels (>14 ng/L at baseline or 3 hours after presentation). Mean age: 63.9±13.3 years	70 (50)	69 (50)	Yes	Yes	Routine ICA versus second generation dual source scanner	ICA during initial hospitalisation
Linde et al Denmark23	VERDICT Observational prospective cohort n=1023 67% male	Substudy of patients with NSTEMI receiving CTCA before randomisation into VERDICT clinical trial, which evaluated outcome of patients with NSTEACS randomised 1:1 to very early or standard ICA. Mean age: 62±12 years	1023 (100)	n/a	No	No	Routine ICA versus minimum 64-detector scanner	Diagnostic accuracy with an ‘intention to diagnose’ analysis
Hinderks et al Netherlands24	Observational prospective cohort n=20 55% male	Patients presenting with NSTEMI with planned ICA within 72 hours after admission, clinically blinded CTCA performed before. Mean age: 68±11 years	19 (100)	n/a	No	No	Routine ICA with FFR/iFR versus 320-detector scanner with myocardial CT perfusion imaging (CT-MPI)	Diagnostic accuracy CTCA and CTA-MPI versus ICA with FFR/iFR
Maffei et al Italy25	Observational retrospective cohort n=1500 79% male	Registry of patients with NSTEMI and no prior revascularisation who received CTCA and ICA. Mean age: 63±10 years	237 (100)	n/a	No	No	ICA versus 64-detector scanner	Diagnostic accuracy
Meier et al France11	Observational prospective cohort n=151 70% male	Patients with high-risk NSTEACS, receiving CTCA with FFR-CT analysis, followed by ICA with invasive FFR. Mean age: 63±12 years	151 (100)	n/a	No	No	Routine ICA with invasive FFR versus 256-detector scanner with FFR-CT	Diagnostic accuracy FFR CT compared with CTCA alone
Kuhl et al Denmark26	Observational prospective cohort n=400 76% male	Patients with NSTEMI examined with CTCA prior to ICA. Mean age: 61±12 years	370 (100)	n/a	No	No	Routine ICA versus 64-detector scanner	Diagnostic accuracy for stenosis ≥50%

ACS, Acute coronary syndrome; BYPASS-CTCA, Randomised Controlled Trial to Assess Whether Computed Tomography Cardiac Angiography Can Improve Invasive Angiography in Bypass Surgery; CTCA, CT coronary angiography; FFR, Fractional flow reserve; hs-cTnT, high sensitivity cardiac troponin T assay; ICA, invasive coronary angiography; iFR, instantaneous wave-free ratio; MPI, myocardial perfusion imaging; NSTEACS, Non-ST elevation acute coronary syndrome; NSTEMI, non-ST-segment-elevation myocardial infarction; RAPID-CTCA, Rapid Assessment of Potential Ischaemic Heart Disease with CTCA; RCT, randomised controlled trial; SoC, Standard of care; VERDICT, Very Early Versus Deferred Invasive Evaluation Using Computerized Tomography in Patients With Acute Coronary Syndromes.

Use of ICA after CTCA

The role of initial CARdiovascular Magnetic rEsonance imaging and computed Tomography Angiography (CARMENTA) in non-ST-elevation myocardial infarction patients¹⁰ and RAPID-CTCA (combined n=896 patients with NSTEMI)²² used findings from CTCA to guide use of ICA (figure 2). By contrast, the protocol for BYPASS-CTCA (n=310 patients with NSTEMI) recommended ICA following CTCA.¹⁷ However, 13 patients did not receive ICA due to clinical decision.

Figure 2. Use of ICA after CTCA, combined clinical outcomes and risk of bias assessments. This figure outlines the use of ICA after CTCA versus the standard-of-care approach in the three randomised controlled trials. The combined clinical outcomes per trial are also shown, along with associated appraisal results with the Risk of Bias 2 Tool. The three studies appear to have low concerns with regard to quality of evidence. CTCA, CT coronary angiography; ICA, invasive coronary angiography; ACE, major adverse cardiovascular events; MI, myocardial infarction.M.

RAPID CTCA reported a relative risk (RR) of 0.99 (0.92–1.08) for the use of ICA during index admission for patients in their CTCA group compared with the standard-of-care group. CARMENTA reported a RR of 0.66 (0.55–0.78) for use of ICA. A random-effects meta-analysis with the inverse-variance method with DerSimonian-Laird estimation was performed with these studies (online supplemental appendix figure 1). This showed that a CTCA first approach compared with standard care was associated with a lower RR of ICA use (pooled RR=0.82 (0.55–1.22), I²=94.6%). However, these two studies exhibited high heterogeneity.

Combined clinical outcome for MI, all-cause death and cardiovascular death

There were ‘zero events’ in CARMENTA for all-cause death and in view of the heterogeneity of trial design across RAPID CTCA, BYPASS-CTCA and CARMENTA (figure 2), a pooled RR for clinical outcomes may misrepresent data and was therefore not performed (online supplemental appendix figure 2). All reported outcomes are at 1 year. RAPID CTCA reported a RR of 0.93 (0.56–1.55) for all-cause death and acute MI. CARMENTA reported a RR of 0.82 (0.26–2.57) for MACE (all-cause death, acute MI and new heart failure). BYPASS-CTCA reported an RR of 0.63 (0.43–0.94) for all-cause death and acute MI.

Diagnostic accuracy of CTCA compared with ICA

The five observational studies (n=1800) reported diagnostic accuracy (figure 3). This information was not reported in the three RCTs. The aggregate unadjusted measures for patient-level diagnostic accuracy of CTCA versus ICA for patients with >50% coronary artery diameter stenosis are shown in figure 4; accuracy: 90.5% (89.2%–91.9%), positive likelihood ratio: 4.09 (2.62, 5.56), negative likelihood ratio: 0.04 (0.03, 0.05). A random effects meta-analysis was conducted, and the pooled sensitivity of CTCA was 96.3% (90.0%–98.7%) (I²=75.7%) while the pooled specificity was 79.6% (59.9%–91.1%) (I²=87.3%). Heterogeneity was substantial for both values indicating large variability between studies. The pooled area under the curve was estimated by a random-effects meta-analysis at 0.88 (0.80–0.96). Substantial interstudy heterogeneity was noted (I²=95.5%, p<0.001).

Figure 3. Diagnostic accuracy, hospital length of stay, quality of life and patient satisfaction and risk of bias assessments. This figure provides a combined illustration of true positives (TP), false positives (FP), true negatives (TN) and false negatives (FN) regarding diagnostic accuracy data provided by observational trials with associated criteria and risk of bias assessment with the QUADAS-2 tool. Length of stay, quality of life and patient satisfaction data are also presented here as documented from included records. CTCA, CT coronary angiography; EuroQol VAS, Euroqol visual analogue scale; ICA, invasive coronary angiography.

Figure 4. Forest plot of aggregate sensitivity, specificity, PPV and NPV for coronary artery diameter stenosis >50%. This forest plot shows aggregate data regarding sensitivity, specificity, PPV and NPV as presented by each observational trial along with the combined aggregate unadjusted estimation. CTCA, CT coronary angiography; ICA, invasive coronary angiography; NPV, negative predictive value; NSTEMI, non-ST-segment-elevation myocardial infarction; PPV, positive predictive value.

Procedure-related complications

Data on procedure-related complications in a CTCA-first versus the standard care approach were only identified in the BYPASS-CTCA study. Procedural-related complications were noted to be reduced with a reported OR of 0.08 (0.02–0.34), p=0.001.

Hospital length of stay, patient satisfaction and healthcare costs

Although some studies reported the stand-alone costs, patient satisfaction and length of stay of either CTCA or ICA, there was a lack of comparative studies which report these outcomes for patients who received CTCA with follow-on ICA and ICA alone in the same analysis. Figure 3 displays the extracted data for these outcome measures. Pooled mean length of stay of patients undergoing a CTCA-first approach was 5.37 (±8.47) days, while for standard of care it was 5.67 (±8.07) days as meta-analysed from RAPID CTCA and CARMENTA (n=828). RAPID-CTCA and BYPASS-CTCA reported patient satisfaction on a five-point Likert scale and European Quality of Life (EuroQoL). It was not possible to calculate pooled patient satisfaction scores as BYPASS CTCA did not allow discretionary use of ICA. However, both studies described positive patient satisfaction and quality of life metrics in their CTCA-first approaches. There was no information on associated costs of a CTCA-first approach in comparison to standard care in patients with an established diagnosis of NSTEMI.

Study quality

We assessed the risk of bias with the appropriate tools based on study design. This included RoB 2 for RCTs, ROBINS-I for non-randomised intervention studies and QUADAS-Comparative for diagnostic accuracy studies. GRADE was used to appraise study quality. A summary of risk-of-bias is presented in figures 2 and 3 while a summary of GRADE study quality per analysed outcome is presented in figure 5. The overall quality of evidence and risk of bias was variable with noticeable limitations surrounding areas such as confounding and imprecision.

Figure 5. GRADE quality of evidence assessment for analysed outcomes from randomised controlled and diagnostic accuracy studies. This figure illustrates the GRADE quality of evidence assessment for analysed outcomes from both randomised controlled trials and observational studies. There is moderate to low certainty due to the limited evidence available. BYPASS-CTCA, Randomised Controlled Trial to Assess Whether Computed Tomography Cardiac Angiography Can Improve Invasive Angiography in Bypass Surgery; CARMENTA, CARdiovascular Magnetic rEsonance imaging and computed Tomography Angiography; CTCA, CT coronary angiography; MI, myocardial infarction; RAPID-CTCA, Rapid Assessment of Potential Ischaemic Heart Disease with CTCA.

Among the three RCTs, there appeared to be low risk of bias in most domains. CARMENTA was limited by a small sample size. We have noted heterogeneity between protocols in these studies with variations in outcome reporting and non-discretionary use of ICA in BYPASS-CTCA. From figure 5, the methodological quality of the three RCTs was good. However, certainty was moderate for all studies due to issues with small sample sizes and large 95% CI with small effect sizes.

The non-randomised observational studies showed more variation. From the ROBINS-I assessment, the studies by Hinderks et al, Kühl et al, Linde et al and Maffei et al had moderate to serious concerns in multiple domains23,26 (online supplemental appendix table 1). Confounding, outcome measurement and participant selection were the most common problems. In contrast, the study by Meier et al was assessed as low risk of bias across all domains with a clearer study design.¹¹ Diagnostic accuracy was evaluated using QUADAS-C (figure 3). Most had moderate risk of bias in key areas such as patient selection, index test and reference standard. Meier et al raised concerns with early termination of the trial due to COVID-19, limited sample size and 8% of patients excluded due to image quality despite good quality methodology.¹¹ A consistent limitation across all studies was failure to identify or adjust for confounding factors and the lack of precision in outcome reporting. Publication bias was identified as a concern for Hinderks et al and Maffei et al.^{24 25} This was due to the lack of pre-registration, single centre design for Hinderks et al and speculative conclusions despite small sample sizes.²⁴ These contributed to low evidence quality in figure 5.

Discussion

We have conducted, to our knowledge, the first systematic review and meta-analysis of CTCA diagnostic accuracy and use to guide subsequent therapy in patients diagnosed and treated for NSTEMI. Analysis of pooled data from two RCTs investigating upstream CTCA prior to ICA in patients with NSTEMI found no difference between groups in subsequent use of ICA. Upfront use of CTCA did not alter clinical outcomes in RAPID-CTCA and CARMENTA. By contrast, BYPASS-CTCA reported a significant reduction in subsequent MI along with all-cause death if CTCA was used alongside ICA. The change in this prespecified secondary outcome may reflect the use of CTCA to improve procedure planning and decision making in patients with NSTEMI and more complex post coronary artery bypass grafting anatomy.

Differences in design between the RCTs impacted subsequent utilisation of ICA. RAPID-CTCA used CTCA to facilitate early diagnosis and treatment of patients with undifferentiated intermediate risk acute chest pain. Use of CTCA to direct subsequent ICA was allowed but not embedded in the RAPID-CTCA protocol.²⁷ The smaller CARMENTA Trial recruited patients with NSTEMI who had been scheduled for ICA, and consequently, all patients in the standard-of-care arm received this investigation compared with a corresponding 77% in RAPID-CTCA. Information from CTCA was used to direct subsequent ICA in CARMENTA and a significant 34% reduction in subsequent ICA was observed. The potential for CTCA to reduce ICA use in patients with NSTEMI was further demonstrated from pooled analysis of the single arm observational studies. First, in keeping with previous studies of patients with NSTEMI,^{28 29} 32% (n=568) of patients had normal or non-obstructive disease (maximum stenosis <50%). Second, CTCA had excellent diagnostic accuracy in this population, comparable to performance in patients with stable coronary disease,⁵ with 96% sensitivity and 92% negative predictive value for detecting obstructive coronary artery stenosis >50%. However, specificity (76%) and positive predictive value (90%) for CTCA were lower and variable across the observational studies, compared with values observed in patients with stable disease. This may relate to challenges with overdiagnosis in patients with calcific disease giving rise to blooming artefacts. The potential for current CTCA technology to reduce ICA use in patients with NSTEMI will be demonstrated with imminent completion of the ongoing trial, the Effect of Computed Tomography on Rates of Invasive Coronary Angiography in Acute Coronary Syndrome (ACTRN12621000224820).

CTCA use resulted in a reduction in length of stay (0.3 days/7.2 hours) from pooled RAPID-CTCA and CARMENTA data. This result differs from the slight increased length of stay (approximately 5 hours) associated with upfront CTCA for all-comers in RAPID-CTCA.¹⁶ Data from the RCTs on other secondary outcomes including patient quality of life, patient satisfaction and healthcare costs were unsuitable for meta-analysis owing to heterogeneity in measurement and protocol differences. For example, in BYPASS-CTCA, routine follow-on ICA was recommended, preventing inclusion of this study in meta-analysis of CTCA-guided ICA use. Differences in procedural complications were not reported in the RAPID-CTCA and CARMENTA trials.¹⁶ However, BYPASS-CTCA did reach its primary outcomes of improved clinical outcomes with reduced contrast nephropathy, reduced procedure time, reduced procedural complications and improved patient satisfaction with upstream CTCA.¹⁷ A reduction in procedure-related complications was also observed in the DISCHARGE trial with a CTCA-first approach in patients with stable coronary artery disease.⁹

Strengths of the meta-analysis include its scope in exploring the use of CTCA in patients with NSTEMI from different geographical regions and systematic methodology to identify potential confounders across studies. There are limitations. We noted a high degree of heterogeneity with a limited number of RCTs that fitted the inclusion criteria. Many older studies failed to differentiate between non-differentiated high-risk chest pain including unstable angina and NSTEMI, reducing availability for inclusion and limiting available data for meta-analysis. There was low standardisation between study protocols. Heterogeneity surrounding pooled analyses for diagnostic accuracy was high (I²>50%) while analyses of clinical outcomes and use of ICA were constrained by ‘zero-event’ rates and small pre-existing sample sizes reducing the accuracy of the pooled estimates. A key issue that was highlighted was the low availability of data that focused solely on patients with NSTEMI. Finally, presenting an unadjusted pooled analysis of diagnostic accuracy from observational cohort studies may tend to overestimate the diagnostic accuracy of CTCA in this setting. For example, it is likely that diagnostic accuracy may be lower in older patients with risk markers for coronary calcification which can limit diagnostic interpretation.

Despite good evidence for diagnostic accuracy and safety, this systematic review highlights the need for more research examining CTCA-guided management of patients with NSTEMI. The potential to reduce the use of ICA would be expected to lead to a reduction in procedure-related complications.⁹ The positive impact on patient experience from reduced length of stay and avoidance of an invasive procedure, together with the possible improvement in clinical outcomes resulting from improved revascularisation planning, needs further assessment.

Delivering ICA within guideline-recommended timelines for patients with NSTEMI is challenging for patients who are admitted to hospitals without invasive facilities. A delay for inter-hospital transfer extends the length of hospital stay and contributes to bed pressure. Our data indicate that CTCA is safe and would be effective at selecting patients for ICA. In conclusion, the current volume of evidence examining clinical outcomes of a CTCA-first approach in patients with NSTEMI is limited. Further research with large scale clinical trials is required to reveal potential improvements in clinical outcomes, patient satisfaction and resource utilisation of this novel pathway compared with the current standard of care.