Coronary CT Angiography: The Trials That Actually Changed Practice
Coronary CT angiography moved from a throughput tool in the emergency department to a Class 1, Level A first-line test for stable chest pain in under fifteen years. The arc is unusual: the technology won not by finding more stenoses to stent, but by reclassifying diagnoses and driving preventive therapy. Revascularization rates in SCOT-HEART were identical at ten years — 15.2% vs 15.3% — and the myocardial infarction curves still separated.
What follows is the practice-changing set, grouped by the question each trial was built to answer. For each: the question, the finding, and what changed afterwards — guideline class and level, FDA action, or CPT code. Where nothing changed, that is stated plainly, because the neutral trials (BEACON, RAPID-CTCA, FORECAST, TARGET-CTCA) have shaped the field as much as the positive ones. A regulatory timeline and a master summary table close the piece.
1. Acute Chest Pain in the Emergency Department
Question: In 699 low-risk ED patients (TIMI <4, normal ECG and enzymes) across 16 US sites, does CCTA beat rest–stress SPECT MPI on time and cost to diagnosis?
Finding: Time to diagnosis 2.9 h vs 6.3 h (54% reduction, p<0.0001); ED costs 38% lower ($2,137 vs $3,458, p<0.0001); radiation 11.5 vs 12.8 mSv (p=0.02, favouring CCTA, since the MPI arm received rest and stress). MACE after a normal index test 0.8% vs 0.4% (p=0.29) — underpowered.
What changed: Nothing immediately — but it entered the Level A evidence pool for the 2021 AHA/ACC Chest Pain Guideline and established CCTA as a triage instrument rather than an outcomes instrument.
Question: In 1,370 ED patients with TIMI 0–2 randomized 2:1, is a negative CCTA safe enough to discharge directly from the ED? Primary endpoint was safety — cardiac death or MI at 30 days among CCTA-negative patients.
Finding: Of 640 patients with a negative CCTA, zero died or sustained MI at 30 days (95% CI 0–0.57). ED discharge 49.6% vs 22.7% (difference 26.8 points, 95% CI 21.4–32.2); median stay 18.0 vs 24.8 h (p<0.001); CAD detection higher (9.0% vs 3.5%).
What changed: This is the trial that made “CCTA-negative equals safe discharge” a defensible disposition. Cited directly in the 2021 Chest Pain Guideline (Class 1, LOE A for CCTA in intermediate-risk acute chest pain) and in the 2022 SCCT expert consensus on CCTA in the ED.
Question: In 1,000 patients aged 40–74 with non-ischemic ECG and negative initial troponin, does early CCTA shorten length of hospital stay?
Finding: LOS 23.2 vs 30.8 h (−7.6 h, p<0.001); direct ED discharge 47% vs 12% (p<0.001); no missed ACS in either arm. But more downstream testing (p<0.001), higher cumulative radiation (13.9 vs 4.7 mSv), and no cost saving ($4,289 vs $4,060, p=0.65). ACS prevalence was only 8%.
What changed: It supplied the caveat language that guidelines still carry — efficiency yes, with downstream testing and radiation as the price. It is the empirical basis for the guideline warning that CCTA increases catheterization and revascularization without reducing hard events.
Question: In 562 low-to-intermediate-risk ED patients after a negative initial troponin, how does dual-source CCTA compare with exercise stress ECG for diagnostic accuracy, stay, and cost?
Finding: ACS occurred in 24 patients (4%). CCTA sensitivity 100% (95% CI 81.5–100), specificity 94% versus exercise ECG sensitivity 83% (95% CI 36–99.6), specificity 91%. LOS 13.5 vs 19.7 h (p<0.0005), costs AUD $2,193 vs $2,704 (p<0.001) — offset by more downstream testing (10.8% vs 5.8%, p=0.020) and more invasive angiography (7.1% vs 3.3%, p=0.028).
What changed: Part of the evidence that displaced exercise ECG as a front-line ED strategy; the 2021 US guideline no longer supports routine exercise ECG in acute chest pain. Nothing changed on this trial alone.
Question: In 500 Dutch ED patients with suspected ACS managed with high-sensitivity troponin, does early CCTA increase identification of significant CAD requiring revascularization at 30 days?
Finding: Negative. Revascularization 9% vs 7% (p=0.40); length of stay identical at 6.3 h (p=0.80); no increase in direct discharge. The only gains were lower direct medical cost (€337 vs €511, p<0.01) and less outpatient testing (4% vs 10%, p<0.01).
What changed: Once an efficient hs-cTn rule-out pathway exists, CCTA's throughput advantage evaporates. A principal reason European guidelines never granted CCTA a Class I in the acute setting.
Question: In 1,748 higher-risk suspected ACS patients across 37 UK hospitals (mean GRACE 115; elevated troponin, ECG changes, or known CAD), does early CCTA reduce all-cause death or type 1/4b MI at 1 year?
Finding: Negative — 5.8% vs 6.1%, adjusted HR 0.91 (95% CI 0.62–1.35), p=0.65. Invasive angiography fell (54.0% vs 60.8%, HR 0.81, 0.72–0.92, p=0.001); length of stay rose by 0.21 days; preventive prescribing was unchanged.
What changed: The 2023 ESC ACS Guideline downgraded CCTA from Class I (2020) to Class IIa, Level A, restricted to non-elevated or uncertain hs-cTn without ECG change or recurrent pain. No guideline recommends routine CCTA in troponin-positive ACS.
Question: In 404 participants, does quantitative plaque burden predict 1-year death or non-fatal MI better than the GRACE score or obstructive stenosis?
Finding: Above-median low-attenuation plaque burden: HR 7.80 (95% CI 2.33–26.0), p<0.001 — outperforming GRACE >140 (HR 3.80) and obstructive disease (HR 2.07, p=0.065, non-significant). Event patients had LAP burden 4.22% vs 2.14% (p<0.001).
What changed: Nothing formally. It is the principal acute-setting argument for phenotyping plaque rather than grading stenosis, and is reflected in the 2025 ACC statement on quantitative plaque analysis — but no class of recommendation exists.
Question: In 3,170 ED patients across 14 UK hospitals in whom MI had already been ruled out but hs-cTnI indicated intermediate risk (peak >5 ng/L), does outpatient CCTA-guided care reduce MI or cardiac death?
Finding: Negative at median 3.0 years — 7.1% vs 7.3%, adjusted HR 0.95 (95% CI 0.73–1.23), p=0.71 — despite statins in 63.4% vs 48.5%, antiplatelets in 42.0% vs 32.8%, and more PCI (6.2% vs 4.9%). Crossover was negligible. Obstructive CAD was found in 22.6%, non-obstructive in 42.5%.
What changed: Nothing yet. It directly pressures the US Class 1, LOE A acute recommendation and vindicates the ESC's Class IIa. The mechanistic message is that troponin elevation in this population is largely not coronary in origin — expect this to be the pivotal citation at the next chest pain guideline revision.
Question: Across every randomized comparison of CCTA versus standard care in acute chest pain, what does CCTA actually deliver?
Finding: MI RR 0.86 (95% CI 0.66–1.12); all-cause mortality RR 0.96 (0.59–1.58); invasive angiography RR 1.08 (0.89–1.30); revascularization RR 1.37 (1.08–1.74) — the only statistically significant outcome. Length of stay fell ~14% overall, and radiation rose by a mean of 7.24 mSv in higher-risk patients.
What changed: Nothing directly, but it is the honest summary statement for the ED literature: CCTA buys efficiency and more revascularization, not fewer events. ACS accounts for only 5–10% of ED chest pain presentations, which constrains any achievable absolute benefit.
2. Stable Chest Pain — Anatomy Displaces Function
Question: In 4,146 patients at 12 Scottish chest pain clinics with suspected stable angina, does adding CCTA to standard care change the certainty of diagnosis — and eventually, the events?
Finding: Diagnostic certainty RR 1.79 (1.62–1.96); the CHD diagnosis was reclassified in 27% and angina in 23%, versus 1% with standard care. At 5 years: CHD death or non-fatal MI 2.3% vs 3.9%, HR 0.59 (0.41–0.84), p=0.004, with no increase in invasive angiography (23.7% vs 24.2%) and more preventive therapy (OR 1.40, 1.19–1.65). At 10 years: 6.6% vs 8.2%, HR 0.79 (0.63–0.99), p=0.044; non-fatal MI HR 0.72; revascularization identical (15.2% vs 15.3%).
What changed: The mechanism is statins, not stents. SCOT-HEART underpins NICE CG95 (2016) making CTCA first-line, the Class 1, LOE A recommendation in the 2021 US guideline, and Class I, Level A in the 2024 ESC CCS Guideline.
Question: In 10,003 symptomatic outpatients (52.7% women; mean pretest likelihood 53.3%), is anatomic-first CCTA superior to functional testing for death, MI, unstable angina hospitalization, or major procedural complication?
Finding: No difference — 3.3% vs 3.0%, adjusted HR 1.04 (0.83–1.29), p=0.75. Catheterization without obstructive CAD was lower (3.4% vs 4.3%, p=0.02) but catheterization overall was higher (12.2% vs 8.1%). At 10.6 years, mortality was 14.3% vs 14.5%, HR 0.98 (0.87–1.10).
What changed: Established CCTA as an equivalent, guideline-acceptable first test — the basis of the “either/or” Class I wording in the 2019 ESC CCS Guideline. The long-term null is the strongest counterweight to over-claiming a survival benefit for anatomic testing.
Question: Does the CAD-RADS category carry prognostic weight, and can a clinical model identify patients who need no test at all?
Finding: In 3,840 patients, event rates rose stepwise: CAD-RADS 0 = 0.8%, 2 = 4.3%, 4a = 10.2%, 4b/5 = 18.5%, with adjusted HR up to 21.84 (8.63–55.26). The c-statistic was 0.747 for CAD-RADS versus 0.657 for calcium score and 0.629 for the ASCVD risk score. The PROMISE Minimal Risk Tool classified 26.8% of patients as minimal risk (c-index 0.725).
What changed: Converted CAD-RADS from a reporting convention into a risk stratifier, and supplied the model later deployed prospectively in PRECISE and DEFER-CTA. Note a published correction to the minimal-risk tool paper for coding errors.
Question: In 3,561 European patients (56.2% women) with stable chest pain and intermediate pretest probability (10–60%) already referred for invasive angiography, is CT a safe initial strategy?
Finding: MACE over 3.5 years 2.1% vs 3.0%, HR 0.70 (0.46–1.07) — no significant difference — but major procedure-related complications 0.5% vs 1.9%, HR 0.26 (0.13–0.55). Obstructive CAD was present in only ~25% of those referred to the cath lab.
What changed: Cemented CT as a gatekeeper before invasive angiography; cited for the Class I, Level A CCTA recommendation at >5–50% clinical likelihood in the 2024 ESC CCS Guideline.
Question: CONSERVE: in 1,611 patients already booked for ICA, is selective referral after CCTA non-inferior? CRESCENT: does a calcium-score-gated tiered CT protocol beat functional testing? CAPP: does CT beat exercise ECG on patient-reported outcomes?
Finding: CONSERVE — events 4.6% in both arms (p=0.026 for non-inferiority), yet only 23% vs 89% underwent ICA and the normalcy rate fell from 61% to 25%. CRESCENT — event-free survival 96.7% vs 89.8% (p=0.011), downstream testing 25% vs 53% (p<0.0001), diagnostic cost €369 vs €440. CAPP — better Seattle Angina Questionnaire stability (difference −11.1, p=0.001) and inconclusive results in 2.4% vs 26.9%.
What changed: Validated the CAC-gated selective CCTA pathway now embedded in the 2024 ESC algorithm (calcium scoring for reclassification at 5–15% likelihood), the “CT before the cath lab” posture of both major guidelines, and NICE's removal of exercise ECG from the diagnostic pathway.
Question: In 2,103 stable symptomatic patients, does a precision strategy — defer testing in minimal-risk patients, otherwise CCTA with selective FFR-CT — beat usual testing on death, MI, or catheterization without obstructive CAD?
Finding: Composite 4.2% vs 11.3%, HR 0.35 (0.25–0.50), p<0.001 (covariate-adjusted HR 0.29); catheterization without obstructive CAD 2.6% vs 10.2%. In the deferred 20%: composite 0.9% vs 6.3%, adjusted HR 0.16 (0.04–0.70), with zero deaths and zero MIs, and frequent angina falling from ~70% to <20% at 12 months even in untested patients. Caveat: non-fatal MI was numerically higher in the precision arm (1.2% vs 0.5%, HR 2.65, 0.96–7.36) — absence of evidence rather than evidence of absence.
What changed: The evidence base for the risk-factor-weighted clinical likelihood model and the “defer testing at ≤5%” rule in the 2024 ESC CCS Guideline, and for the expanded role of FFR-CT (COR 2a, LOE B-NR in the 2021 US guideline). Subtraction, not addition, is the new frontier.
Question: Can blinded, core-lab-read CCTA exclude left main stenosis ≥50% and confirm obstructive CAD before randomization, replacing pre-randomization invasive angiography?
Finding: 5,757 screening CCTAs; 7.5% excluded for left main disease and 21% for absence of obstructive CAD. Against invasive angiography: 97.1% concordance for excluding left main, 92.2% for identifying at least single-vessel disease, with 4.9% false negatives.
What changed: Proved CCTA can serve as an anatomic gatekeeper at scale — now the standard design for chronic coronary syndrome trials, and part of the Level A basis for CCTA's Class 1 status.
3. CT Physiology — FFR-CT and Perfusion
Question: Can computational fluid dynamics applied to a standard CCTA dataset reproduce invasive FFR ≤0.80?
Finding: DISCOVER-FLOW (103 patients, 159 vessels) — per-vessel accuracy 84.3% versus 58.5% for stenosis alone, with CT specificity of only 39.6%. DeFACTO missed its primary endpoint: per-patient accuracy 73% (95% CI 67–78), the lower bound falling below the prespecified 70%. Algorithm revision to v1.4 produced NXT: per-patient sensitivity 86%, specificity 79%, AUC 0.90 (0.87–0.94) versus 0.81 for CT stenosis, whose specificity was 34%.
What changed: NXT is the trial that earned the FDA De Novo grant DEN130045 (26 November 2014), creating an entirely new device classification, 21 CFR 870.1415. DeFACTO is the instructive failure: the first-generation algorithm was not good enough, and the field said so.
Question: Does FFR-CT prevent invasive angiograms that find nothing, and is deferral on the basis of FFR-CT >0.80 safe?
Finding: PLATFORM — ICA with no obstructive CAD in 12% vs 73% (risk difference 61%, 95% CI 53–69, p<0.0001); 61% of planned invasive studies were cancelled. ADVANCE (n=4,737, 38 sites) — management reclassified in 66.9%; ICA showing no obstructive CAD 14.4% when FFR-CT ≤0.80 versus 43.8% when >0.80; and zero MACE among 1,592 patients with FFR-CT >0.80 at 90 days.
What changed: NICE MTG32 (February 2017) recommended HeartFlow FFR-CT for stable recent-onset chest pain, with modelled savings of £214 per patient, revised to £391 in 2021. These data feed the COR 2a, LOE B-NR FFR-CT recommendations for 40–90% proximal or mid stenoses in the 2021 US Chest Pain Guideline. PLATFORM's weakness is its consecutive-cohort, non-randomized design.
Question: In randomized designs, does FFR-CT save money (FORECAST), work as an on-site machine-learning tool (TARGET), scale to 5,297 patients (Radiology 2024), and reduce unnecessary ICA without industry funding (FUSION)?
Finding: FORECAST was negative on its primary endpoint — total 9-month cardiac costs £1,605 vs £1,491, p=0.10 — though ICA fell 19% vs 25% (p=0.01). TARGET (n=1,216, China) — the composite of ICA without obstructive disease or failure to revascularize obstructive disease: 28.3% vs 46.2%, p<0.001, with 1-year MACE unchanged. The Radiology 2024 trial (n=5,297, 17 centres) — ICA within 90 days 10.0% vs 12.4%, p=0.006. FUSION (n=528, funded by Zorginstituut Nederland) — unnecessary ICA at 90 days 18% vs 33%, p<0.001, sustained at 1 year (22% vs 39%), with revascularization identical at 20% (p=0.948).
What changed: FFR-CT gained Category I CPT 75580 effective 1 January 2024 (replacing Category III 0501T–0504T, in place since 2018), with Medicare LCD coverage restricted to 40–90% stenosis and explicitly not alongside stress testing. Guideline status remains split: COR 2a (US, 2021) versus Class IIb (ESC, 2024). FUSION is the first independent randomized confirmation and is the likely lever for an upgrade.
Question: Does adding stress CT perfusion to CCTA improve detection of flow-limiting disease, and how does it compare head-to-head with FFR-CT?
Finding: CORE320 (381 patients, 8 countries) — AUC 0.87 (0.84–0.91) vs 0.84 for CCTA alone, p=0.02: significant but clinically modest. PERFECTION — accuracy 92% (CCTA + stress CTP) vs 87% (CCTA + FFR-CT), essentially equivalent. A 2024 head-to-head against invasive FFR found no difference (AUC 0.84 vs 0.83). DECIDE-Gold published only a design paper; primary results never appeared.
What changed: Nothing. CTP has no recommendation in the 2021 US Chest Pain Guideline, no standalone recommendation in the 2024 ESC CCS Guideline, and no Category I CPT code. Comparable accuracy was not enough to overcome the stress agent, the second acquisition, the added radiation, and the absence of outcome trials. This is the cleanest “nothing formally changed” story in cardiac CT.
4. From Stenosis to Plaque — and to AI
Question: Does non-obstructive CAD carry prognostic risk — and what did the culprit segment look like on a CCTA performed years before the event?
Finding: CONFIRM (n=23,854, 27 centres) — CCTA findings independently predicted all-cause mortality, rising stepwise with the number of obstructive vessels. ICONIC (234 ACS cases propensity-matched 1:1) — 75% of culprit precursors were <50% stenosis at baseline and only 4.65% were ≥70%; fibrofatty plus necrotic core volume 65.2 vs 45.6 mm³ (p=0.008); high-risk plaque in 52% vs 33.3% (p=0.003).
What changed: No class or level anywhere — but these are the datasets that broke the stenosis-centric model and justified plaque-based reporting. Nearly every subsequent plaque study is built on the CONFIRM substrate.
Question: Do positive remodeling, low-attenuation plaque, and the napkin-ring sign predict events independent of stenosis severity and risk scores?
Finding: PROMISE (n=4,415; HRP in 15.1%) — adjusted HR 1.72 (1.13–2.62) overall, and 4.31 (2.25–8.26) in non-obstructive disease, with the strongest signal in women (2.41) and younger patients (2.33). SCOT-HEART (n=1,769, median 4.7 years) — adverse plaque HR 3.01 (1.61–5.63); obstructive plus adverse plaque versus normal coronaries HR 11.50 (3.39–39.04); quantitative low-attenuation plaque burden >4% conferred HR 4.65 (2.06–10.5) for MI, outperforming calcium score and stenosis severity.
What changed: Incorporated as modifiers in CAD-RADS 2.0 (2022) and in the 2024 SCCT quantitative standards. No guideline makes high-risk plaque a treatment trigger — it remains descriptive, and the 4% LAP threshold has no endorsement as an action threshold.
Question: Do statins, icosapent ethyl, or colchicine measurably change plaque on serial CCTA?
Finding: PARADIGM (n=1,255, mean interval 3.8 ± 1.6 years) — statins slowed total atheroma progression by 21%, increased calcified (+1.27 vs +0.98%/year, p<0.001) and reduced non-calcified (+0.49 vs +1.06%/year, p<0.001) progression, with 35% less high-risk-plaque development — while stenosis progression was unchanged. EVAPORATE (n=80, 18 months) — low-attenuation plaque −17% vs +109%, p=0.0061. Colchicine (n=80, non-randomized) — LAP volume −40.9% vs −17.0%, p=0.039 adjusted.
What changed: Calcification as stabilization entered the clinical vocabulary. None of these changed a guideline: icosapent ethyl labeling rests on REDUCE-IT, colchicine's Class 2b recommendation and the 2023 FDA approval rest on LoDoCo2 and COLCOT, and the 2025 ACC statement calls serial plaque analysis inadequately supported. A related caution: PREVENT (preventive PCI of vulnerable plaque, 2-year composite 0.4% vs 3.4%, HR 0.11) enrolled by intracoronary imaging, not CCTA — it is not CT evidence.
Question: Does the perivascular fat attenuation index predict cardiac death beyond conventional CCTA — including in the majority of patients sent home with no obstructive CAD?
Finding: CRISP-CT (derivation n=1,872; validation n=2,040) — FAI ≥−70.1 HU: cardiac mortality HR 9.04 (3.35–24.40) and 5.62 (2.90–10.88). ORFAN (n=40,091) — in non-obstructive CAD, per-1-SD FAI cardiac death HR 1.60–1.86 by vessel; three inflamed vessels versus none, HR 29.8 (13.9–63.9); an AI-risk category of “very high” conferred HR 6.75 for cardiac death.
What changed: FDA De Novo DEN250042 granted to CaRi-Heart on 28 July 2026 — a new classification (21 CFR 870.2215) for a predictive indicator of long-term cardiovascular outcomes, and the first authorized technology to quantify coronary inflammation from a routine CCTA. No CPT code and no guideline recommendation for FAI as yet.
Question: Does AI-QCT match expert readers, change what physicians actually do, and predict events better than a trained human?
Finding: CLARIFY (n=232) — CAD-RADS agreement within one category 98.3% (weighted κ 0.812); stenosis ≥70% accuracy 99.1%. CERTAIN (n=750, crossover) — diagnosis or management changed in 57.1% (p<0.001), downstream testing −37.1%, statin initiation or intensification +28.1%, aspirin initiation +23.0%. CONFIRM2 — prognostic AUC rose from 0.67 (risk factors) to 0.77 with AI-QCT (p<0.001), with 3-year event rates of 1.3% / 3.2% / 9.9% across risk strata; head-to-head, MACE AUC 0.81 (AI-QCT) vs 0.79 (CAD-RADS), p<0.001 and 0.81 vs 0.70 vs calcium score, with net reclassification improvement of 0.47–0.61.
What changed: Multiple FDA clearances (Cleerly, HeartFlow K250902, Elucid K241524, Artrya, Circle, Caristo, Keya), and Category I CPT 75577 for coronary plaque assessment effective 1 January 2026, replacing Category III 0623T–0626T. Yet the 2025 ACC Scientific Statement on quantitative coronary plaque analysis declines to endorse routine use, restricts it to patients with visually evident plaque, names no plaque-volume threshold, and calls the significance of small AI-detected volumes unknown. TRANSFORM (>7,000 patients, staged plaque-directed therapy) has not reported. Reimbursement is ahead of the evidence.
5. CT as a Procedural Roadmap
Question: Does cross-sectional MDCT annular sizing outperform 2D TEE and reduce paravalvular regurgitation?
Finding: Moderate-or-worse PVL 21.9% (TEE-guided) vs 7.5% (CT-guided), p=0.045, with TEE measurements non-discriminatory (CT ΔDmax AUC 0.82, p<0.001). In the prospective multicentre sizing-algorithm study (n=266): more-than-mild paravalvular AR 5.3% vs 12.8%, p=0.032 and severe 0% vs 4.5%, p=0.013.
What changed: CT replaced TEE as the sizing standard of care worldwide within roughly two years — yet the 2020 ACC/AHA Valvular Heart Disease Guideline carries no Class or LOE recommendation for it, mentioning CT only in a preprocedural testing table and supporting text. The normative document is the 2019 SCCT TAVI consensus. Practice changed faster than the guideline text.
Question: Can surgeons plan and perform CABG for left main or multivessel disease using CCTA plus FFR-CT alone, with no invasive angiogram at all?
Finding: Single-arm, multicentre proof of concept, n=114 — feasibility 99.1% (95% CI 95.2–100) of operations completed on CT guidance without recourse to invasive angiography, with 30-day anastomosis patency 92.6%.
What changed: Nothing yet — single-arm, low-risk, n=114. But it is the first credible demonstration that the diagnostic catheterization may become optional before surgery. A randomized successor is required before this can enter a guideline.
Question: Does the FFR-CT Planner — virtual stenting performed on the CT dataset — predict post-PCI invasive FFR?
Finding: In 120 patients with motorized FFR pullback and OCT before and after PCI: predicted versus measured post-PCI FFR differed by 0.02 ± 0.07 (limits of agreement −0.12 to +0.15); across 97,402 matched pullback values, the mean difference was 0.01 ± 0.05. Predicted minimal stent area 5.0 ± 2.2 mm² versus OCT-measured 5.60 ± 2.01 mm².
What changed: The 2025 SCAI/SCCT roundtable endorses CCTA for case triage, complexity assessment, shared decision-making, chronic total occlusion and post-CABG patients, and FFR-CT-based virtual PCI — while stating plainly that CCTA-based stent sizing lacks prospective validation, with no randomized comparison against intravascular-imaging-guided PCI. P4, the randomized trial of CT-guided versus angiography-guided PCI, has not reported.
Question: How accurate is CCTA for bypass graft patency and in-stent restenosis, and what is its guideline standing before AF ablation and LAA occlusion?
Finding: Graft meta-analysis (50 studies, 3,449 patients, 7,506 grafts) — pooled sensitivity 0.96, specificity 0.97, NPV 0.98, overall accuracy 0.97, independent of graft type. In-stent restenosis remains limited by blooming and beam-hardening, with acceptable accuracy mainly in stents ≥3.0 mm. For LAA imaging, delayed-phase acquisition raises thrombus specificity to 98–100%.
What changed: Graft imaging is the least controversial CT indication. But pre-ablation cardiac CT appears in neither the 2023 ACC/AHA/ACCP/HRS nor the 2024 ESC atrial fibrillation guidelines, and the 2025 SCAI/HRS LAAO guideline suggests pre-procedure TEE or CT — a conditional recommendation on very low certainty evidence, treating the two as interchangeable. Near-universal practice, essentially no guideline text.
1. High-risk plaque has no class of recommendation anywhere. Low-attenuation plaque burden outperformed the GRACE score and obstructive disease for predicting death or MI — yet no guideline tells you what to do differently when you see it.
2. AI plaque quantification is reimbursed before it is recommended. Cleared by the FDA, coded (CPT 75577), paid by Medicare, and covered by most commercial lives — while the ACC's own 2025 statement restricts it to select scenarios and publishes no threshold. TRANSFORM is the trial that will settle it.
3. Universal practice, absent guideline text. CT for TAVR sizing, pre-ablation pulmonary vein anatomy, and LAA occlusion planning are near-universal — yet TAVR sizing has no Class or LOE in the 2020 valvular guideline, pre-ablation CT appears in neither AF guideline, and LAAO treats CT and TEE as interchangeable on very low certainty.
6. Regulatory and Reimbursement Timeline
| Date | Action | Significance |
|---|---|---|
| Nov 2014 | FDA De Novo DEN130045 — HeartFlow FFR-CT | Created a new device class, 21 CFR 870.1415. Earned by NXT. |
| Aug 2016 | NICE CG95 update | CTCA first-line; pretest probability tables and exercise ECG abolished. |
| Feb 2017 | NICE MTG32 — FFR-CT | Adoption supported; £214–391 modelled saving per patient. |
| Jan 2018 | CPT Category III 0501T–0504T | First tracking codes for FFR-CT. |
| 2019–2020 | Cleerly K191802, K202280; HeartFlow Planner | AI plaque quantification enters the US market. |
| Oct 2021 | AHA/ACC Chest Pain Guideline | CCTA Class 1, LOE A; FFR-CT COR 2a, LOE B-NR for 40–90% stenoses. |
| Oct 2022 | HeartFlow K213857 — Plaque Analysis + RoadMap | Plaque characterization added to a cleared physiology platform. |
| Aug 2023 | 2023 ESC ACS Guideline | CCTA in acute chest pain downgraded Class I → IIa, after RAPID-CTCA. |
| Jan 2024 | CPT Category I 75580 — FFR-CT | Permanent code; Category III codes deleted. Petitioned by ACC, ACR and SCCT. |
| Aug 2024 | 2024 ESC CCS Guideline | CCTA Class I, Level A at >5–50% likelihood; CT-FFR Class IIb. |
| Jan 2025 | CMS OPPS — CCTA reassigned to APC 5572 | Hospital outpatient payment roughly doubled, from ~$175 to ~$357; provisional. |
| Jul 2025 | HeartFlow K250902; Elucid K241524 (Sept 2024) | Next-generation plaque tools; the vendor field broadens. |
| Dec 2025 | ACC Scientific Statement on QCPA | Declines routine use; no threshold; serial analysis called unsupported. |
| Jan 2026 | CPT Category I 75577 — coronary plaque assessment | Replaces 0623T–0626T. AI plaque analysis becomes an established service. |
| Jul 2026 | FDA De Novo DEN250042 — CaRi-Heart | First authorized quantification of coronary inflammation (21 CFR 870.2215). |
| Aug–Sep 2026 | FUSION and TARGET-CTCA report | FFR-CT strengthened; CCTA in post-rule-out acute chest pain challenged. |
7. Master Summary Table
| Trial | Year / n | Headline result | What changed |
|---|---|---|---|
| CT-STAT | 2011 / 699 | Diagnosis in 2.9 vs 6.3 h; 38% cheaper | Level A pool for 2021 US guideline |
| ACRIN-PA | 2012 / 1,370 | 0 events in 640 CCTA-negative patients | Class 1, LOE A acute CCTA |
| ROMICAT-II | 2012 / 1,000 | LOS −7.6 h; more testing, no cost saving | Guideline caveat language |
| CT-COMPARE | 2014 / 562 | Sensitivity 100% vs 83% for exercise ECG | Helped retire exercise ECG |
| BEACON | 2016 / 500 | Neutral; LOS identical with hs-cTn | No Class I in Europe |
| RAPID-CTCA | 2021 / 1,748 | Neutral; HR 0.91 (0.62–1.35) | ESC 2023 downgrade to IIa |
| TARGET-CTCA | 2026 / 3,170 | Neutral; HR 0.95 despite more statins | Pending; challenges Class 1 |
| SCOT-HEART | 2015–25 / 4,146 | 10-y HR 0.79; revascularization identical | NICE 2016; Class 1 A both sides |
| PROMISE | 2015–25 / 10,003 | Equivalent; 10.6-y mortality HR 0.98 | “Either/or” Class I wording |
| DISCHARGE | 2022 / 3,561 | Complications 0.5% vs 1.9%, HR 0.26 | ESC 2024 Class I A gatekeeper |
| CONSERVE | 2019 / 1,611 | ICA 23% vs 89%, events identical | “CT before the cath lab” |
| PRECISE / DEFER-CTA | 2023 / 2,103 | HR 0.35; deferral safe in 20% | ESC 2024 defer-at-5% rule |
| NXT | 2014 / 254 | AUC 0.90 vs 0.81; specificity 79% vs 34% | FDA De Novo DEN130045 |
| PLATFORM | 2015 / 584 | Normal ICA 12% vs 73% | NICE MTG32 |
| ADVANCE | 2018 / 4,737 | Reclassified 66.9%; 0 MACE if >0.80 | COR 2a FFR-CT, 2021 |
| FORECAST | 2021 / 1,400 | Neutral on cost (p=0.10) | Nothing; a useful corrective |
| FUSION | 2026 / 528 | Unnecessary ICA 18% vs 33% | Pending; likely 2a→upgrade lever |
| CORE320 | 2014 / 381 | AUC 0.87 vs 0.84 (p=0.02) | Nothing — CTP never entered guidelines |
| ICONIC | 2018 / 468 | 75% of culprits <50% at baseline | Broke the stenosis model |
| SCOT-HEART LAP | 2020 / 1,769 | LAP >4%: HR 4.65 for MI | CAD-RADS 2.0 modifier only |
| ORFAN | 2024 / 40,091 | 3 inflamed vessels: HR 29.8 | FDA De Novo DEN250042 |
| CERTAIN / CONFIRM2 | 2024–26 | Management changed 57%; AUC 0.81 vs 0.79 | CPT 75577, Jan 2026 |
| CT annular sizing | 2012–13 | Moderate PVL 21.9% → 7.5% | Practice standard; no Class/LOE |
| FASTTRACK CABG | 2024 / 114 | 99.1% feasible without ICA | Nothing yet; RCT required |
| P3 | 2022 / 120 | Post-PCI FFR predicted to 0.02 ± 0.07 | SCAI/SCCT 2025 endorsement |
Anatomic-first testing won on diagnostic yield, procedural safety, and preventive therapy — not on revascularization and not on all-cause survival. The guideline sequence reads cleanly: NICE 2016 → ESC 2019 (Class I, Level B) → AHA/ACC 2021 (Class 1, Level A, above stress imaging at Level B) → ESC 2024 (Class I, Level A, explicitly first-line at >5–50% likelihood).
The 2026 frontier is subtraction — deferring testing in minimal-risk patients, and using FFR-CT to keep patients out of the cath lab — while the plaque and inflammation era arrives with FDA clearances and CPT codes running ahead of randomized outcome data. TRANSFORM, P4, and the next chest pain guideline revision are where this gets settled.
8. References — stable DOI links
Two naming cautions for readers: TARGET (Circulation 2023, on-site CT-FFR in China) is a different trial from TARGET-CTCA (NEJM 2026, post-rule-out acute chest pain); and CPT 75580 is FFR-CT while CPT 75577 is coronary plaque assessment — the two are frequently conflated in secondary reporting. The 2026 citations (TARGET-CTCA, FUSION, the CaRi-Heart De Novo, CPT 75577) were confirmed against publisher and regulatory records at the time of writing; DOIs for congress late-breakers should be re-checked once fully indexed and swapped for the live reference if the article identifier changes.
