Background: Invasive coronary angiography (ICA) is performed in many patients with low-risk non-ST-elevation acute coronary syndrome (NSTE-ACS), yet a substantial share have no obstructive coronary disease. Coronary CT angiography (CTCA) could triage these patients, but prospective data from India are scarce.Objectives: To determine the diagnostic accuracy of CTCA for significant coronary artery disease (CAD, ≥50% stenosis) against ICA in haemodynamically stable patients with unstable angina (UA) or NSTEMI and a low TIMI risk score (0–2); and to assess agreement, the effect of coronary calcification, potential ICA avoidability and in-hospital outcomes.Methods: In this prospective single-centre study at S.M.S. Medical College, Jaipur, 160 patients (UA 72, NSTEMI 88) underwent CTCA (≥64-slice MDCT with Agatston scoring) followed by ICA 24–72 hours later. ICA was the reference standard. Patient-level sensitivity, specificity, predictive values, ROC area, Cohen's kappa and McNemar's test were calculated; vessel-level agreement, calcium-score and diagnosis subgroups and in-hospital major adverse cardiac events (MACE) were analysed.Results: Mean age was 58.4 ± 11.2 years; 71.3% were men. ICA showed significant CAD in 116 patients (72.5%). CTCA had sensitivity 94.8% (95% CI 89.1–98.1), specificity 90.9% (78.8–97.5), PPV 96.5%, NPV 87.0% (73.7–95.1), accuracy 93.8% and AUC 0.929 (0.876–0.965). Patient-level kappa was 0.84 (0.76–0.92) and vessel-level kappa 0.86; McNemar p = 0.754. Accuracy fell from 100% (Agatston 0) to 84.2% (Agatston>400). All six false negatives had Agatston>400. Accuracy was similar in UA and NSTEMI (91.7% vs 95.5%). CTCA showed no significant CAD in 46 patients (28.8%), of whom six had disease on ICA. In-hospital MACE occurred in 12 patients (7.5%) and was associated with CTCA-defined triple-vessel disease, left main involvement and higher calcium score. Effective dose was lower for CTCA than ICA (4.8 ± 1.4 vs 6.2 ± 2.1 mSv).Conclusions: In stable, low-risk UA/NSTEMI, CTCA showed high accuracy and near-perfect agreement with ICA. It could safely defer ICA in roughly one patient in four, provided heart rate is controlled and the calcium burden is low; in heavily calcified coronaries (Agatston>400) a negative or equivocal CTCA should not on its own replace ICA.
Coronary artery disease remains the leading cause of cardiovascular death worldwide, with a rising burden in low- and middle-income countries including India.[1] Unstable angina (UA) and non-ST-elevation myocardial infarction (NSTEMI) together form the non-ST-elevation acute coronary syndromes (NSTE-ACS), which account for a growing share of admissions to tertiary cardiac centres.[2] Current guidelines recommend invasive coronary angiography (ICA), with an early invasive strategy for high-risk patients.[2]
ICA gives definitive anatomy and allows same-sitting revascularisation, but it is invasive and carries risks of access-site complications, contrast nephropathy, coronary dissection and radiation exposure. Reported series show that a considerable proportion of patients sent for ICA with suspected NSTE-ACS have no obstructive disease, especially at low clinical risk, so some invasive procedures may be avoidable with sound non-invasive triage.
Coronary CT angiography (CTCA) provides sub-millimetre, three-dimensional imaging of the coronary tree with a high negative predictive value for obstructive stenosis. The ESC 2019 chronic coronary syndrome guideline recommends it as a first-line anatomical test in suitable patients.[3] In stable chest pain, the DISCHARGE trial found CT-first and ICA-first strategies similar for cardiovascular events with fewer procedural complications in the CT arm,[4] and the 10-year SCOT-HEART results showed fewer coronary death or non-fatal myocardial infarction events with CTCA-guided care.[5] A meta-analysis of randomised trials likewise supported CTCA-first management in stable chest pain.[6]
Evidence in the acute setting is thinner. Early 64-slice CTCA reliably excluded significant disease in symptomatic patients but tended to overestimate stenosis severity.[7] The VERDICT study evaluated CTCA before ICA in NSTE-ACS,[8] and its prognostic analysis found that CTCA-defined disease carried prognostic information comparable to ICA.[9] More recent work in suspected NSTEMI reports high sensitivity but modest specificity, with calcification a recognised source of false positives.[10,11] Few prospective data exist from India, where patients are younger at presentation, have a heavy metabolic risk burden and often show extensive coronary calcification, all of which may affect CTCA performance.
We therefore prospectively compared CTCA with ICA in haemodynamically stable patients with UA or NSTEMI and a low TIMI risk score. Our primary aim was the patient-level diagnostic accuracy of CTCA for ≥50% stenosis. Secondary aims were agreement with ICA, the effect of coronary calcification, the proportion of patients who might avoid ICA, and in-hospital clinical outcomes.
Study design and setting
This was a prospective, single-centre, observational diagnostic accuracy study in the Department of Cardiology, S.M.S. Medical College and Attached Group of Hospitals, Jaipur, a large government tertiary referral centre in Rajasthan, over six months [insert start and end dates]. It followed the Declaration of Helsinki, was approved by the Institutional Ethics Committee, and all participants gave written informed consent in English or Hindi. The report follows STARD 2015.
Participants
Consecutive adults (≥18 years) admitted with UA or NSTEMI (clinical presentation, ECG, and high-sensitivity cardiac troponin above the 99th percentile for NSTEMI), a TIMI risk score of 0–2, and haemodynamic stability (systolic pressure ≥90 mmHg, no cardiogenic shock, no refractory chest pain) were eligible. Exclusion criteria were STEMI or haemodynamic instability; known severe CAD or prior CABG; prior coronary stenting; eGFR<30 mL/min/1.73 m²; contrast allergy; uncontrolled arrhythmia; pregnancy; and TIMI score ≥3.
Sample size
Sample size was estimated from the sensitivity formula n = Z² × Sn × (1 − Sn) / d², using an expected sensitivity of 92% from a published meta-analysis,[6] 95% confidence and 7% absolute precision. This gave about 58 patients with disease. A minimum of 116 patients was set, and 160 were enrolled to allow for dropouts and non-diagnostic scans.
Clinical and laboratory assessment
We recorded history, risk factors (diabetes, hypertension, dyslipidaemia, smoking, family history, obesity), admission haemodynamics and a 12-lead ECG. The TIMI score was calculated at enrolment.[12] Laboratory tests were high-sensitivity troponin I/T, serum creatinine with eGFR, and a fasting lipid profile.
CT coronary angiography
CTCA used a ≥64-slice multidetector scanner with ECG gating. Prospective triggering was used where feasible (heart rate ≤65 bpm). Intravenous metoprolol was given if resting heart rate exceeded 65 bpm, and sublingual nitroglycerin immediately before acquisition. A non-contrast Agatstoncalcium score was obtained[13] and grouped as 0, 1–100, 101–400 and >400. An experienced cardiologist blinded to clinical data read the images. Each segment was graded normal, non-obstructive (<50%) or significant (≥50%), and the four territories (left main, LAD, LCX, RCA) were reported separately. Significant CAD meant ≥50% stenosis in any major territory.
Invasive coronary angiography (reference standard)
ICA was done by the standard Judkins technique via femoral or radial access, at least 24 hours and within 72 hours after CTCA, with multiple projections for each territory. Stenosis was graded visually, and significant CAD was again ≥50% stenosis in any major territory. Revascularisation decisions followed standard clinical practice.
Outcomes
The primary outcome was patient-level sensitivity, specificity, PPV, NPV and accuracy of CTCA against ICA. Secondary outcomes were patient- and vessel-level agreement (Cohen's kappa), ROC analysis, severity-grade concordance, subgroup performance by Agatston category and by UA versus NSTEMI, the proportion with no significant CAD on CTCA (potential ICA avoidability), management strategy, in-hospital MACE (death, myocardial infarction, urgent revascularisation) and radiation dose.
Statistical analysis
Continuous data are mean ± SD and categorical data n (%). Diagnostic indices were derived from the 2 × 2 table with 95% confidence intervals. Agreement used Cohen's kappa (Landis and Koch categories[14]) and McNemar's test; ROC analysis gave the AUC and Youden index. Subgroups were analysed by Agatston category and diagnosis. A two-sided p < 0.05 was significant.
A total of 160 patients (72 UA, 88 NSTEMI) were analysed. The mean age was 58.4 ± 11.2 years (range 32–82) and 114 (71.3%) were men. All had a TIMI score of 0–2. Hypertension, dyslipidaemia and diabetes were the commonest risk factors (Table 1). All CTCA-to-ICA intervals fell within the 24–72-hour window (mean 38.4 ± 10.8 hours), and prospective ECG-gating was used in 72% of scans.
Table 1. Baseline characteristics (N = 160)
|
Characteristic |
Value |
|
Age, years, mean ± SD |
58.4 ± 11.2 |
|
Male sex, n (%) |
114 (71.3) |
|
Diagnosis: UA / NSTEMI, n (%) |
72 (45.0) / 88 (55.0) |
|
TIMI score 0 / 1 / 2, n (%) |
38 (23.8) / 68 (42.5) / 54 (33.7) |
|
Hypertension, n (%) |
102 (63.8) |
|
Diabetes mellitus, n (%) |
84 (52.5) |
|
Dyslipidaemia, n (%) |
96 (60.0) |
|
Current smoking, n (%) |
72 (45.0) |
|
Family history of CAD, n (%) |
48 (30.0) |
|
Obesity (BMI ≥30 kg/m²), n (%) |
52 (32.5) |
|
Normal admission ECG, n (%) |
26 (16.2) |
|
eGFR, mL/min/1.73 m², mean ± SD |
78.6 ± 14.2 |
|
LDL cholesterol, mg/dL, mean ± SD |
126.8 ± 32.7 |
|
HDL cholesterol, mg/dL, mean ± SD |
38.2 ± 7.4 |
|
Agatston score 0 / 1–100 / 101–400 / >400, n (%) |
22 (13.7) / 44 (27.5) / 56 (35.0) / 38 (23.8) |
CTCA showed significant CAD in 114 patients (71.2%), non-obstructive disease in 32 (20.0%) and normal coronaries in 14 (8.8%). ICA confirmed significant CAD in 116 patients (72.5%); of the 44 without it, 14 had normal coronaries and 30 non-obstructive disease. On both modalities the LAD was the most often affected vessel (≥50% stenosis in 56.2%), followed by the RCA (42.5%), LCX (36.2%) and left main (12.5%). On ICA, 56 patients (35.0%) had single-vessel, 38 (23.8%) double-vessel and 22 (13.7%) triple-vessel disease.
At patient level, CTCA gave 110 true positives, 4 false positives, 6 false negatives and 40 true negatives (Table 2). Sensitivity was 94.8% and specificity 90.9% (Table 3). The AUC was 0.929 (SE 0.022, p < 0.001) and the Youden index 0.857, with ≥50% stenosis as the operating threshold. McNemar's test showed no directional bias (p = 0.754).
Table 2. CTCA versus ICA for significant CAD (≥50%), patient level
|
|
ICA positive |
ICA negative |
Total |
|
CTCA positive |
110 |
4 |
114 |
|
CTCA negative |
6 |
40 |
46 |
|
Total |
116 |
44 |
160 |
Table 3. Diagnostic performance of CTCA (ICA as reference)
|
Measure |
Value |
95% CI |
|
Sensitivity |
94.8% |
89.1–98.1 |
|
Specificity |
90.9% |
78.8–97.5 |
|
Positive predictive value |
96.5% |
91.3–99.0 |
|
Negative predictive value |
87.0% |
73.7–95.1 |
|
Accuracy |
93.8% |
88.8–97.0 |
|
AUC |
0.929 |
0.876–0.965 |
|
Cohen's kappa (patient level) |
0.84 |
0.76–0.92 |
Patient-level agreement was 93.8% (kappa 0.84, near-perfect). Across 640 vessels (four per patient), 604 were concordant (94.4%; kappa 0.86). Kappa ranged from 0.83 (LCX) to 0.88 (LAD) (Table 4). By stenosis grade, concordance was highest for normal vessels (100%) and 70–99% stenosis (95.8%) and lowest for 25–49% stenosis (87.5%).
Table 4. Vessel-level agreement between CTCA and ICA (640 vessels)
|
Vessel |
Concordant (n) |
Discordant (n) |
Agreement (%) |
Kappa |
|
Left main |
154 |
6 |
96.3 |
0.84 |
|
LAD |
152 |
8 |
95.0 |
0.88 |
|
LCX |
148 |
12 |
92.5 |
0.83 |
|
RCA |
150 |
10 |
93.8 |
0.85 |
|
All vessels |
604 |
36 |
94.4 |
0.86 |
Diagnostic performance declined steadily as calcium burden rose: accuracy was 100% at Agatston 0 and 84.2% (specificity 75.0%) above 400 (Table 5). All six false-negative scans had Agatston>400 and five (83.3%) were acquired at a heart rate above 75 bpm; the RCA was missed in four. Three of the four false positives involved the LCX and three had Agatston>400 (Table 6). Performance did not differ significantly between UA and NSTEMI (accuracy 91.7% vs 95.5%, all p > 0.05; Table 7).
Table 5. CTCA performance by Agatston score
|
Agatston score |
n |
Sensitivity (%) |
Specificity (%) |
Accuracy (%) |
|
0 |
22 |
100 |
100 |
100 |
|
1–100 |
44 |
97.2 |
94.4 |
96.4 |
|
101–400 |
56 |
95.1 |
91.7 |
93.8 |
|
>400 |
38 |
88.0 |
75.0 |
84.2 |
Table 6. Characteristics of discordant cases
|
Feature |
False negative (n = 6) |
False positive (n = 4) |
|
Age, years, mean ± SD |
64.2 ± 8.4 |
52.6 ± 11.2 |
|
Male sex |
83.3% |
50.0% |
|
Diabetes |
66.7% |
25.0% |
|
Agatston>400 |
100% |
75.0% |
|
Heart rate >75 bpm at scan |
83.3% |
25.0% |
|
Vessel most often involved |
RCA (4/6) |
LCX (3/4) |
Table 7. CTCA performance in UA versus NSTEMI
|
Measure |
UA (n = 72) |
NSTEMI (n = 88) |
p |
|
Sensitivity (%) |
92.5 |
96.3 |
0.38 |
|
Specificity (%) |
88.2 |
93.3 |
0.41 |
|
PPV (%) |
94.6 |
97.8 |
0.29 |
|
NPV (%) |
84.2 |
89.5 |
0.44 |
|
Accuracy (%) |
91.7 |
95.5 |
0.31 |
CTCA showed no significant CAD in 46 patients (28.8%): 14 with normal coronaries and 32 with non-obstructive disease. Of these, 40 were confirmed disease-free on ICA and 6 (13.0% of CTCA-negative scans) had significant CAD. Management after ICA was PCI in 82 (51.3%), medical therapy alone in 44 (27.5%), CABG in 24 (15.0%) and combined PCI plus medical therapy in 10 (6.3%).
In-hospital MACE occurred in 12 patients (7.5%): death in 2 (1.3%), myocardial infarction in 4 (2.5%) and urgent revascularisation in 6 (3.7%). Patients with MACE more often had CTCA-defined triple-vessel disease, left main involvement and a higher Agatston score (Table 8). With only 12 events, these are unadjusted associations.
Table 8.CTCA findings by in-hospital MACE
|
CTCA finding |
MACE (n = 12) |
No MACE (n = 148) |
p |
|
Significant CAD |
100% |
68.2% |
0.018 |
|
Triple-vessel disease |
58.3% |
10.1% |
<0.001 |
|
Left main involvement |
41.7% |
9.5% |
0.002 |
|
Agatston score, mean ± SD |
486.4 ± 124.8 |
188.6 ± 142.2 |
<0.001 |
|
TIMI score 2 |
75.0% |
30.4% |
0.003 |
Mean effective dose was 4.8 ± 1.4 mSv for CTCA (range 2.6–8.2) and 6.2 ± 2.1 mSv for ICA (range 3.4–12.6) (p < 0.01). ICA fluoroscopy time averaged 8.6 ± 3.2 minutes.
In this prospective study of haemodynamically stable, low-risk UA/NSTEMI, CTCA identified significant CAD with a sensitivity of 94.8%, specificity of 90.9% and overall accuracy of 93.8%, with near-perfect patient- and vessel-level agreement with ICA. Accuracy was unaffected by diagnosis (UA vs NSTEMI) but fell with coronary calcification, and all missed lesions occurred at Agatston scores above 400.
Comparison with previous work
Our sensitivity is close to that of the multicentre 64-slice study of Meijboom et al. (99% sensitivity, 64% specificity, 97% NPV in 360 symptomatic patients)[7] and of the NSTEMI cohort of Van Cauteren et al. (95% sensitivity, 65% specificity for CAD-RADS ≥3 in 66 patients).[11] Our specificity was higher than in both. Plausible reasons are newer scanners, heart-rate control with prospective gating in 72% of scans, and a low-risk case mix with fewer heavily calcified vessels; the exclusion of non-diagnostic scans may also have contributed. Wang et al. conclude that CTCA is an effective gatekeeper to the catheterisation laboratory in stable NSTEMI but that data on outcomes remain limited,[10] and the VERDICT investigators found CTCA-defined disease to have prognostic value comparable to ICA-defined disease in NSTE-ACS.[8,9] Our MACE analysis, though small, is consistent with this: left main involvement, triple-vessel disease and a high calcium score on CTCA were associated with in-hospital events.
Our findings also fit the stable chest pain literature, in which CT-first strategies matched ICA-first strategies for cardiovascular events with fewer procedural complications in DISCHARGE,[4] and were associated with fewer coronary deaths or myocardial infarctions at 10 years in SCOT-HEART.[5] Those trials enrolled stable patients, so they support but do not establish the use of CTCA-first pathways in NSTE-ACS.
Calcification, heart rate and false results
The fall in accuracy from 100% at Agatston 0 to 84.2% above 400, with specificity of 75.0% in the latter group, matches the known blooming effect of dense calcium on stenosis grading. This matters in a population with a heavy metabolic burden and frequent calcification. The six false negatives were all in heavily calcified vessels, mostly scanned at heart rates above 75 bpm and mainly in the RCA, a vessel prone to motion artefact. The false positives were mainly LCX lesions in calcified segments. The practical message is that image quality and calcium score should be part of the decision to defer ICA, and that heart-rate control (target below 65 bpm) is worth the effort.
Clinical implications and a cautious reading of ICA avoidability
CTCA showed no significant disease in 46 of 160 patients (28.8%), and in 40 of these (25.0% of the cohort) ICA was confirmed to be unnecessary for diagnosis. Because every patient had ICA by design, this is a potential rather than observed reduction in invasive procedures, and its size (roughly one in four) is in line with the reduction in diagnostic ICA reported for CT-first strategies in stable chest pain.[6] However, our NPV of 87.0% (95% CI 73.7–95.1) is below the 90% or more often quoted, and 13% of CTCA-negative patients had significant CAD. A deferral policy should therefore be limited to diagnostic-quality scans with low calcium burden and adequate heart-rate control, and patients with a high Agatston score should go to ICA or to functional testing, such as CT-derived fractional flow reserve, before ICA is withheld. Where resources are limited, as in many public-sector tertiary centres in India, such a pathway could ease catheterisationlaboratory load. Effective dose was lower with CTCA than with ICA in our series, although CTCA is an addition to, not a substitute for, ICA unless ICA is actually avoided.
Strengths and limitations
In haemodynamically stable patients with low-risk UA or NSTEMI, CTCA showed high diagnostic accuracy and near-perfect agreement with ICA. Roughly one in four patients had no significant disease on CTCA and might avoid ICA, provided scans are of diagnostic quality, heart rate is controlled and calcium burden is low. In heavily calcified coronaries (Agatston above 400), CTCA alone should not replace ICA. Randomised and longer-term data are needed before CTCA-first pathways are adopted routinely in this population.