Failure of Standard Contemporary STEMI ECG Criteria to Identify Acute LAD Occlusion

Authors, Journal, Affiliations, Type, DOI

• Authors: H. Pendell Meyers, Scott W. Sharkey, Robert Herman, José Nunes de Alencar, Gautam R. Shroff, William H. Frick, Stephen W. Smith
• Journal: European Heart Journal: Acute Cardiovascular Care (2025), Vol. 14, pp. 403–411
• Affiliations: Carolinas Medical Center (Charlotte, NC); Minneapolis Heart Institute; University of Naples Federico II; Dante Pazzanese Institute of Cardiology (São Paulo); Hennepin Healthcare/University of Minnesota; SSM Health St. Louis University Hospital
• Type: Retrospective sub-study / case-control (DOMI-ARIGATO database, ClinicalTrials.gov NCT03863327)
• DOI: https://doi.org/10.1093/ehjacc/zuaf037
• Conflicts: H.P.M., R.H., and S.W.S. own stock in and/or consult for Powerful Medical (manufacturer of PMCardio Queen of Hearts AI ECG model)

Overview

This retrospective sub-study of the DOMI-ARIGATO case-control database (808 patients, 265 with OMI) evaluated the sensitivity of standard STEMI ECG criteria (4th Universal Definition of MI) for detecting total LAD occlusion (TIMI-0 flow) — the highest-risk OMI territory. Among 53 confirmed LAD TIMI-0 cases, 38% (20/53) never met STEMI criteria on any serial pre-angiography ECG despite multiple ECGs in 16/20 cases. Both a blinded expert interpreter (S.W.S.) and the PMCardio Queen of Hearts AI ECG model achieved 100% sensitivity on the very first ECG in all 53 cases, including all 20 missed by STEMI criteria. Despite equivalent infarct sizes, patients without STEMI criteria had nearly 2.5× longer door-to-balloon times (97 vs 40 min, P<0.001), underscoring the clinical cost of missing these cases under the current STEMI-activation paradigm.

Keywords

Acute coronary syndromes, STEMI, OMI, LAD occlusion, STEMI criteria, Electrocardiography

Key Takeaways

Study Design and Cohort

• Sub-study of DOMI-ARIGATO: 808 patients with symptoms of possible ACS presenting to two academic EDs (Stony Brook University Hospital and Hennepin County Medical Center)
• Primary analysis cohort: 53 patients with LAD culprit artery and confirmed TIMI-0 flow at angiography
• All serial ECGs were interpreted blindly by expert (S.W.S.) for OMI vs not-OMI; all first ECGs were run through the PMCardio Queen of Hearts AI model
• STEMI criteria applied per 4th Universal Definition of MI: ≥1 mm STE in all leads except V2–V3 (≥2 mm men ≥40y; ≥2.5 mm men <40y; ≥1.5 mm women)
• Inter-rater agreement for STEMI criteria: 97.2% (kappa 0.893); for OMI diagnosis: 94.0% (kappa 0.849)

STEMI Criteria Sensitivity for LAD TIMI-0 OMI

• 38% (20/53) of total LAD occlusions never met STEMI criteria on ANY serial ECG before angiography
• 16 of 20 had at least 2 pre-angiography ECGs; median time first-to-last ECG was 44 min (IQR 13–147 min) — none evolved to STEMI criteria
• For broader TIMI-0 occlusions (any artery, n=148): STEMI criteria sensitivity was only 49%
• Context: published meta-analysis documents only 43% sensitivity of traditional STEMI criteria for any acute coronary occlusion

Expert and AI Performance

• Expert (S.W.S.) and AI model (PMCardio Queen of Hearts): 100% sensitivity on the first ECG for all 53 LAD TIMI-0 cases (P<0.0001 vs STEMI criteria, χ²)
• Independent blinded cardiologist (W.H.F.) agreed with absence of STEMI criteria in 19/20 cases (95%); one borderline case resolved by computer measurement
• At fixed specificity, AI model demonstrated double sensitivity (67% vs 33%) of STEMI criteria in the validation cohort

ECG Findings in the 20 Missed Cases

• Subtle STE not meeting criteria (<1 mm): 85% (17/20)
• Hyperacute T-waves: 85% (17/20)
• Pathologic Q-waves (associated with subtle STE): 70% (14/20)
• Reciprocal STD and/or T-wave inversion: 50% (10/20)
• Terminal QRS distortion: 20% (4/20)
• Inferior STE with aVL STD/T-inversion: 20% (4/20)
• Posterior OMI pattern (STD maximal V2–V4): 0%
• Modified Sgarbossa criteria (LBBB/VPR): 0%
• Note: Hyperacute T-waves did NOT evolve to STEMI criteria in any case (16/16 with serial ECGs, 94%)

Clinical Outcomes

• Door-to-balloon time (DBT): 97 min (IQR 60–428) in those WITHOUT STEMI criteria vs 40 min (IQR 22–61) in those WITH STEMI criteria (P<0.001)
• DBT >90 min: 55% (11/20) without STEMI criteria vs 21% (7/33) with STEMI criteria (P=0.012)
• Infarct size equivalent: Peak troponin and post-angiography ejection fraction did NOT differ between STEMI(+) and STEMI(−) LAD OMI (P=0.94 for EF; P=0.52–0.53 for troponin)
• EF <50% in 55% of the 20 missed cases; 1 death during index hospitalisation
• LAD occlusion site in the 20 missed cases: proximal (n=8), mid (n=12)
• Discharge diagnosis paradox: Among the 20 STEMI(−) OMI patients, those with DBT <120 min were likely to receive "STEMI" as final diagnosis (9/12, 75%), while those with DBT >120 min were likely to receive "NSTEMI" (7/8, 88%) — suggesting diagnostic labels were influenced by treatment delays rather than actual ECG findings

Implications for the OMI/NOMI Paradigm

• Current STEMI-based model: ECG serves as "gatekeeper" for emergent cathlab activation; any occluded artery without STE criteria is routed as NSTEMI → no immediate cathlabs → longer delays
• The OMI/NOMI paradigm (Meyers, Smith et al.) reframes diagnosis around the underlying pathophysiology (persistent occlusion = imminent infarction) rather than a single ECG finding (STE)
• Even subtle ECG features (hyperacute T-waves, subtle STE <1mm, pathologic Q-waves, reciprocal STD) reliably identify OMI when assessed by expert or AI — without requiring STE threshold criteria
• Recognition of this problem has begun in the 2022 ACC Chest Pain guideline and 2024 SCAI Expert Consensus (acknowledging limited STE sensitivity for occlusion), but neither provides specific criteria for hyperacute T-waves or acute posterior MI patterns

Limitations of the Document

• Retrospective study from two academic centers; results may not be generalizable to community hospitals or EMS settings
• Small cohort: 53 LAD TIMI-0 cases; 20 without STEMI criteria
• Consecutive enrolment not guaranteed — high-risk enriched sample; both sites are cath lab activation/referral centres
• Specificity of expert interpretation and AI model could not be calculated in this sub-analysis (no non-OMI control group for this specific sub-study)
• Expert OMI interpretation was performed by only one of two study experts for the LAD TIMI-0 sub-group
• No long-term outcomes or recovery EF data available; proxy measures used (peak troponin and post-angiography EF)
• Financial conflicts: key authors have equity in Powerful Medical (Queen of Hearts developer)

Key Concepts Mentioned

• concepts/OMI-NOMI-Paradigm — central framework; STEMI/NSTEMI replaced by occlusion vs non-occlusion
• concepts/ST-T-Changes — STEMI criteria inadequacy; hyperacute T-waves; subtle STE; de Winter pattern
• concepts/Hyperacute-T-waves — present in 85% of subtle LAD OMI; not defined by amplitude alone; do not evolve to STEMI criteria

Key Entities Mentioned

• entities/Acute-Coronary-Syndrome — context for STEMI/NSTEMI vs OMI/NOMI paradigm debate
• entities/PMCardio-Queen-of-Hearts — AI ECG model achieving 100% sensitivity for LAD OMI

Wiki Pages Updated

• wiki/sources/failure-stemi-criteria-lad-omi-ehjacc-2025.md — created (this file)
• wiki/sourceindex.md — new entry added
• wiki/wikiindex.md — new entries: OMI-NOMI-Paradigm, Hyperacute-T-waves, PMCardio-Queen-of-Hearts
• wiki/concepts/OMI-NOMI-Paradigm.md — created
• wiki/concepts/Hyperacute-T-waves.md — created
• wiki/concepts/ST-T-Changes.md — updated (OMI ECG findings section added)
• wiki/entities/Acute-Coronary-Syndrome.md — updated (OMI paradigm + contradiction)
• wiki/entities/PMCardio-Queen-of-Hearts.md — created