Ischemic ST‐Segment Depression Maximal in V1–V4 (Versus V5–V6) of Any Amplitude Is Specific for Occlusion Myocardial Infarction (Versus Nonocclusive Ischemia)

This was a planned substudy of the Diagnosis of Occlusion MI and Reperfusion by Interpretation of the Electrocardiogram in Acute Thrombotic Occlusion database, which is a 2‐site collaboration designed to study ECG findings in OMI. Stony Brook University Hospital is a suburban, academic hospital and a regional cardiac catheterization referral center. Hennepin Healthcare Medical Center is an urban academic hospital with a cardiac catheterization laboratory. Both emergency departments (EDs) have >100 000 visits per year. Institutional review board approval was obtained at both sites, and there was no extramural funding. Informed consent was waived by both institutional review boards for this retrospective study.

First, each site accessed the cardiac catheterization laboratory activation database (all urgent and emergent left heart catheterizations during 1 year), which provided both cases (OMI) and controls (without OMI). At Stony Brook University Hospital, we added a previously collected prospective cohort of ED patients who were admitted to the cardiology service with suspected ACS during a 6‐month time period (again contributing both cases and controls). To ensure that the final cohort also contains a substantial number of control patients with abnormal ECGs, we added additional controls from Hennepin Healthcare Medical Center by searching the Use of TROPonin In Acute coronary syndromes (UTROPIA) database²² for patients without OMI but with STE, STD, or T‐wave inversion, one‐third of whom had adjudicated nonocclusion MI. Patients were excluded if there were no ECGs in the electronic medical record or if there was insufficient retrospective information available to determine the primary outcome (the presence or absence of our OMI definition).

Chart review was performed by 4 emergency medicine residents after training with a standardized data coding manual. Primary and senior authors (H.P.M. and S.W.S.) were available for on‐demand questions and retraining as necessary. Demographics, clinical and laboratory results, serial ECGs, and angiographic findings were collected using Research and Electronic Data Capture. All available transfer, prehospital, and study site ECGs were collected for each patient.

ECG interpretation included various predefined ECG findings, objective measurements, and subjective interpretations, and was performed by S.W.S. and H.P.M. blinded to all patient information, except age and sex (necessary for interpretation of STEMI criteria, which are age and sex based). S.W.S. interpretations were used for all analyses of OMI versus not OMI and evaluations of STE criteria. Serial ECGs were interpreted sequentially, and initially blinded to the baseline ECG (when available from a previous visit), then unblinded to that baseline ECG for a separate interpretation. The interpreter could not go back and change an interpretation once entered, just as a clinician must interpret each ECG as it is chronologically recorded in real time. STEMI criteria were defined according to the fourth universal definition of MI, and thus measured in millimeters using the QRS onset (PQ junction) and the J‐point. If any ECG before the angiogram met STEMI criteria, the patient was considered to be STEMI(+); if not, then the patient was considered STEMI(−). Interobserver variation to the nearest 0.5 mm has previously been established in our author group.²³ We assessed interobserver reliability for OMI between H.P.M. and S.W.S. for all cases interpreted by both. Furthermore, all 108 consecutive OMI cases from the prospective cohort were reviewed for STEMI criteria by a cardiologist (J.A.K.) blinded to the outcome and the study goals. Specific ECG measurements and observations included STE meeting STEMI criteria, subtle STE not meeting criteria, hyperacute T waves (including de Winter pattern), reciprocal STD and/or negative hyperacute T waves, STD maximal in V1–V4 indicative of posterior OMI, suspected acute pathologic Q waves (Q waves associated with subtle STE, which cannot be attributed exclusively to old MI), terminal QRS distortion,²⁴ any STE in inferior leads with any STD or T‐wave inversion in lead aVL, and positive modified Sgarbossa criteria for patients with left bundle‐branch block or ventricular paced rhythm.

Furthermore, H.P.M. also reviewed all ECGs from all patients, also blinded to outcome, to specifically evaluate for the presence of any STDmaxV1–4, STDmaxV5–6, or specific lead where STD was maximal, and to note any secondary, or nonischemic, cause of the STDmaxV1–4 or STDmaxV5–6, if applicable (eg, “secondary,” or “nonischemic,” refers to STD that is secondary to an abnormal QRS, such as right bundle‐branch block [RBBB] or paced rhythm). “Any STDmax V1–V4” refers to any STD of any magnitude that is maximal in V1–V4 in any context (even if clearly secondary to the QRS complex, such as in RBBB). STDmaxV1–4 that could not be explained by a nonischemic cause was classified as primary STD, not secondary to an identifiable nonischemic source, and referred to as “suspected ischemic STDmaxV1–4.” There was no minimal magnitude of STD required for our interpreters to diagnose STD (ie, even if the deepest STD was <0.5 mm, for example, the interpreter reported STD). When it was uncertain whether maximal STD was in V4 or V5, the lead with greatest STD in proportion to the QRS complex was chosen.

The diagnosis of OMI was adjudicated by structured chart review. In patients who were determined to not have OMI, the diagnosis of AMI was the following: at Stony Brook University Hospital, it was determined by the final diagnosis on the patient’s record; and at Hennepin Healthcare Medical Center, it was determined by strict adjudication of all clinical data, performed by 3 adjudicators as part of the UTROPIA study. Outcomes used to ascertain the presence of OMI on the ECG cannot be based solely on TIMI flow of the lesion at the time of the angiogram because the state of the artery frequently differs between the time of the ECG and the time of the angiogram. Proven STEMI has an open artery in 19% to 36% of cases, depending on whether it is TIMI −1, −2, or −3 flow. Karwowski et al showed that only 64% of 4581 STEMIs had TIMI 0 flow on angiogram.²⁵ Stone et al found that 72% have TIMI 0 or 1 flow.²⁶ Finally, Cox et al found that 80% had TIMI 0, 1, or 2.²⁷ Thus, ≈20% of true STEMIs have TIMI 3 flow at immediate angiogram. As such, the definition of OMI was reproduced from prior studies,^{23, 28, 29, 30, 31, 32} composed of either (1) “confirmed OMI” on cardiac catheterization (defined as an acute culprit lesion with TIMI 0–2 flow) or (2) “presumed OMI with significant cardiac outcome,” defined as any of the following: (a) acute but nonocclusive (TIMI >2) culprit lesion with highly elevated cardiac troponin (contemporary cardiac troponin T ≥1.0 ng/mL [Roche Diagnostics Elecsys; reference range, ≤0.01 ng/mL] or contemporary cardiac troponin I ≥10.0 ng/mL [Abbott Architect 4th generation; reference range, ≤0.030 ng/mL]); (b) if no angiography, then highly elevated cardiac troponin and a new or presumed new regional wall motion abnormality on echocardiography; or (c) ECG positive for STEMI with death before attempted emergent catheterization. Despite the fact that OMI cannot be based solely on TIMI 0 to 1 flow of the culprit lesion for the reasons explained above, we also presented TIMI 0 to 1 culprit lesions as a dedicated outcome. Formal adjudication was made with all available data, including ECGs, cardiac troponins, echocardiograms, and angiogram results. If TIMI flow was not reported, the cine angiogram was reviewed by a cardiologist (G.R.S.) with experience interpreting angiograms. The definition of “highly elevated” cardiac troponin was chosen previously as the most accurate cutoff for differentiating STEMIs from non–ST‐segment–elevation myocardial infarctions using various cardiac troponin assays,^{33, 34, 35, 36, 37} and has subsequently been internally and externally validated.^{23, 28, 30, 32, 38, 39}

We calculated summary statistics for cases and controls, and diagnostic utility statistics (sensitivity, specificity, and accuracy) for each ECG finding. Interobserver agreement was calculated using κ values for categorical variables. Subject characteristics and outcomes were compared between groups using Mann‐Whitney U or Kruskal–Wallis tests for continuous measurements and Pearson χ² or Fisher exact test for categorical measures. All tests were 2 sided, and statistical significance was accepted at the 0.05 level. Descriptive statistics were performed in Research and Electronic Data Capture, whereas other statistical tests were performed with Microsoft Excel (Version 1905; Redmond, WA).

The primary analysis was the specificity of suspected ischemic STDmaxV1–4 for OMI. Secondary goals included the specificity for OMI of any STDmaxV1–4 (whether classified as “suspected ischemic” or “nonischemic”) morphologic characteristics of posterior OMI (QRS, ST‐segment, and T‐wave morphologic characteristics), comparison of patients with STDmaxV1–4 versus those with STDmaxV4–6, and comparison of patients with STDmaxV1–4 with concomitant STEMI criteria versus those without STEMI criteria.

For the 250 Hennepin Healthcare Medical Center patients interpreted by both S.W.S. and H.P.M., there was 97.2% agreement for the determination of STEMI criteria (κ=0.893), and 94% agreement for the classification of OMI (κ=0.893). Interobserver variation to the nearest 0.5 mm, as well as ST‐T wave morphologic agreement, has been previously established within our author group, including specifically between H.P.M. and S.W.S.^{23, 28, 40, 41, 42}

A total of 147 (18%) patients had any STDmaxV1–4, of whom 29 were (blindly, by ECG only) classified as “nonischemic” (either identical to available baseline ECG or explained by non‐OMI diagnosis). Table 2 shows the clinical outcomes of this group as well as the following subgroups based on ECG interpretation.

A total of 147 (18%) patients had STDmaxV1–4, of whom 135 (92%) underwent angiogram, 116 (79%) had an acute culprit lesion, 102 (69%) met our definition of OMI, 101 (68%) underwent percutaneous coronary intervention (PCI), and 70 (48%) had TIMI 0 or 1 culprit flow before intervention. The specificity, sensitivity, and positive likelihood ratio of any STDmaxV1–4 for OMI, regardless of classification (nonischemic or suspected ischemic), was 91.7%, 38.5%, and 4.64 (see all contingency tables in the online appendix in Table S1).

Of the 147 patients with STDmaxV1–4, 29 were classified as “nonischemic,” either baseline or explained by non‐OMI diagnosis (17 RBBB, 2 ventricular paced rhythm, 3 nonspecific interventricular conduction delay, 2 hypokalemia [both verified true], 1 juvenile T‐wave pattern, 1 left ventricular [LV] aneurysm pattern, and 3 uncategorized but proven identical to prior available baseline ECG). Of these 29 patients with explanations for STDmaxV1–4, 3 had OMI (in other words, the “nonischemic” diagnosis falsely reassured against OMI).

The first case involved a patient with normal QRS morphologic features who experienced LCX OMI simultaneously with severe hypokalemia (K=2.6 mEq/L), whose ECG was interpreted (without any clinical context) by both H.P.M. and S.W.S. as hypokalemia mimicking posterior OMI.

The other 2 cases were patients with RBBB who experienced LCX and D1 OMIs, respectively, whose ECGs were interpreted by both interpreters as RBBB with STDmaxV1–4 suspected to be the normal (appropriately discordant) STDmaxV1–4 of RBBB. Only in retrospect was it recognized that these patients actually have STDmaxV1–4 that is excessively discordant to the positive RBBB R’‐wave, deeper than would be expected for RBBB alone.

A total of 118 (15%) patients had “suspected ischemic” STDmaxV1–4 without an accompanying ECG feature, such as RBBB, that would explain the STDmaxV1–4 as preexisting. Of these 118 patients, 113 (96%) underwent angiogram, 106 (90%) had an acute culprit lesion, 99 (84%) met our definition of OMI, 95 (81%) underwent PCI, and 68 (58%) had TIMI 0 or 1 culprit flow before intervention. The specificity, sensitivity, and positive likelihood ratio for the diagnosis of OMI of suspected ischemic STDmaxV1–4 were 96.5%, 37.4%, and 10.67, and for OMI requiring PCI the specificity, sensitivity, and positive likelihood ratio were 96.0%, 39.7%, and 9.94.

Of the 118 with suspected ischemic STDmaxV1–4, 112 (95%) were interpreted as OMI, of whom 110 (98%) underwent angiogram, 104 (93%) had an acute culprit lesion, 99 (88%) met our definition of OMI, 95 (85%) underwent PCI, and 68 (61%) had TIMI 0 or 1 culprit flow before intervention.

Of the 118 patients, 6 were not called OMI by the interpreter, with reasons including the following: 2 cases (1 with a culprit lesion in the left posterolateral artery): there was no other lead with even subtle evidence of OMI; 1 case with no culprit: no agreement on presence of any STD; 1 case with an LCX culprit: opinion that STD was maximal in V5–V6; 1 case with no culprit: associated T‐wave morphologic feature that led the reader to interpret the STD as chronic; and 1 case with no culprit: atrial fibrillation with rapid ventricular response. The specificity, sensitivity, and positive likelihood ratio for OMI of STDmaxV1–4 interpreted as diagnostic of OMI were 97.6%, 37.4%, and 15.60, respectively.

Of the 112 patients interpreted as OMI, 12 (11%) had triple‐vessel disease or left main ACS, 11 of which were also OMI. Of these 12 patients, 10 were diagnosed with AMI, with the other 2 patients having a slight troponin increase noted during the first few serial troponins (0.00–0.02 ng/mL in one case, and 0.00–0.03 ng/mL in the other case), with no troponins further measured and no formal diagnosis of AMI recorded at discharge. A total of 9 (8%) received coronary artery bypass grafting. A total of 107 of 112 underwent formal echocardiogram; 87 (81%) had a wall motion abnormality, of which 84 (96%) were located in the inferior, posterior, and/or lateral walls.

Only 6 of the 112 had no OMI, no culprit, and no triple‐vessel or left main disease. A total of 5 of these 6 were diagnosed with acute MI, and the remaining patient had normal coronaries on emergent catheterization, without AMI. Thus, only 1 patient of 112 diagnosed with OMI by ECG did not have AMI.

There were 8 patients with STDmaxV1–4 who had posterior leads performed and objectively recorded in the electronic medical record.

Of the 99 OMIs detected by STDmaxV1–4, 47 (47%) had accompanying STEMI criteria in other locations. Culprit lesions included the right coronary artery (53), left circumflex artery (32), posterior descending artery (7), smaller branches capable of supplying the posterior wall, such as first obtuse marginal, second obtuse marginal, ramus intermedius, and left posterolateral (30), and other vessels (10).

A total of 13 patients had STDmaxV1–4 but did not meet the definition of OMI (“false positives”); all 13 underwent angiogram, 7 had culprit lesions (2 proximal LAD, 1 mid‐LAD, 1 LCX, 1 middle right coronary artery, 2 second obtuse marginal, and 1 third diagonal), 6 required PCI, and 1 patient died during index visit.

Table 3 shows the ECG characteristics of various groups. A total of 52 patients had STDmaxV1–4 without concomitant STEMI criteria, but only 12 of 112 patients (11%) had STDmaxV1–4 without any other subtle signs of OMI, especially hyperacute T waves and subtle STE, in other leads; all 12 of these had at least 1 mm of STDmaxV1–4. A total of 22 patients in the STEMI(−) OMI group had STDmaxV1–4 of <1 mm, but all of these had subtle findings of OMI in other locations. In other words, although the findings may be extremely subtle, they are supported by findings in other leads. In summary, all patients with posterior OMI had either (1) at least 1 mm STDmaxV1–4 or (2) other subtle findings of OMI in addition to any STDmaxV1–4 (even if <1 mm).

Of the 99 patients with OMI with suspected ischemic STDmaxV1–4, 47 had concomitant STEMI criteria (STEMI[+] OMI), and the remaining 52 (20% of all 265 OMIs) did not (STEMI[−] OMI); these patients with OMI were completely missed by STEMI criteria but correctly identified by suspected ischemic STDmaxV1–4. Table 4 shows the differences in interventions, peak troponins, and time to catheterization between the 2 categories.

Between the STEMI(+) and STEMI(−) groups, there was no statistical difference between the rates of TIMI 0 or 1 culprit lesion flow, the rates of requiring PCI, or the median peak troponin; the only statistically significant differences found were that the STEMI(−) group had longer delays to catheterization and lower likelihood of receiving catheterization within 90 minutes. Four example patients with OMI diagnosed by suspected ischemic STDmaxV1–4 are shown in Figures 1^, 2^, 3, as well as Figure S1 (supplemental online appendix).

Of the 47 patients with OMI who had STDmaxV1–4 with concomitant STEMI criteria, 30 had both ECG findings simultaneously on the first ECG at arrival. The remaining 17 patients all had STDmaxV1–4 noted earlier than STEMI criteria, with a median (interquartile range) delay of 1.00 (0.42–5.72) hours until STEMI criteria developed on a subsequent serial ECG. Thus, in 69 of 99 patients with OMI (70%), the first or only evidence of OMI was any STDmaxV1–4.

Although the determination of STDmaxV1–4 versus STDmaxV5–6 was dichotomized for all other analyses presented, there were 7 patients with STD for which the interpreter noted that STD was basically equal in leads V4 and V5, making this distinction especially difficult in this small subset. All 7 underwent angiography, 5 had culprit lesions (all LCX), 4 had OMI, and 3 received PCI.

A total of 196 (24.3% of 808) had any objective STDmaxV5–6 (whether classified as ischemic or nonischemic). The interpreter labeled these cases as ischemic (65 [33%]), ischemic and indicative of LAD occlusion pattern (32 [16%]), LV hypertrophy (33 [17%]), ventricular paced rhythm (4 [2%]), LV aneurysm morphologic feature (3 [1.5%]), left bundle‐branch block (32 [16%]), RBBB (3 [1.5%]), nonspecific and mild (19 [9.7%]), and other (5 [2.6%]).

A total of 48 of 196 were identified as OMI by expert interpretation, with 22 meeting STEMI criteria in other leads; thus, there were 174 STDmaxV5–6 without STEMI, and 148 STDmaxV5–6 without any signs of OMI elsewhere on the ECG. Of these 148 patients with STDmaxV5–6 and no signs of OMI, 58 were classified as “ischemic” STD and 90 were classified as “secondary” STD (attributed to LV hypertrophy, left bundle‐branch block, or ventricular paced rhythm). Outcomes of each subset are shown in Table 5. Figure S2 (supplemental appendix) shows the quintessential case of diffuse ischemic STD with STDmaxV5–6, with reciprocal STE in aVR, representing ischemia not meeting our definition of OMI in the setting of known severe 3‐vessel disease and prior coronary artery bypass grafting.

Because OMI comprises only ≈2% to 5% of all ED patients with potential ACS, we had insufficient resources to perform a prospective, consecutive cohort study, and instead we performed a retrospective case‐control study to maximize both the number of patients with OMI and patients with non‐OMI with abnormal ECGs. As a result, our population is a high‐risk cohort with ACS, and findings may not apply to undifferentiated ED patients without ACS strongly suspected clinically.

Our study is retrospective and performed at only 2 centers. Few (only 8) of our patients with STDmaxV1–4 had posterior leads performed and recorded in the electronic medical record. Therefore, we were unable to assess the accuracy of posterior lead criteria in conjunction with STDmaxV1–4. Although this is a limitation, it is somewhat telling that, in our study of 808 patients with 3241 ECGs, it was extremely rare in actual clinical practice to record and document posterior leads, even in patients with anterior STD. This indicates that further education on posterior leads is needed, or standard 12‐lead criteria designed to identify posterior OMI are needed. Our data show that STDmaxV1–4 is an accurate marker of OMI without obtaining posterior leads. More important, there are little data supporting that, when there is ischemic STD in V1–V4, STE in posterior leads helps to differentiate posterior OMI from subendocardial ischemia.⁴³

As are all ECG findings when interpreted by humans, the identification of proportionally maximal STD in leads V1–V4 is subjective, requiring dedicated training and experience to accurately identify. This is an important limitation to external validity. However, the current STEMI criteria are also highly subjective, with notoriously poor interrater reliability,^{44, 45, 46, 47} and never met any external validity standards before becoming the universal approach to ECG ischemia interpretation. The relatively high interrater reliability between our 2 interpreters shows that these skills can be taught and learned with a high level of agreement.