Written by Jesse McLaren, with comments from Smith
A 50-year old patient on the medical wards developed acute chest pain, with an ECG labeled (see computer interpretation at the top) and confirmed as normal by the over-reading cardiologist. What do you think?
(Marqette 12SL algorithm)
There’s normal sinus rhythm, normal conduction, normal axis, normal R wave progression, and normal voltages. Lead aVL jumps out as abnormal because there is a discordant T wave inversion and mild ST depression. This is reciprocal to inferior mild ST elevation and hyperacute T waves (wide based, bulky, and symmetric, and in III taller than the QRS complex), and adjacent to ST depression in V2.
This is diagnostic of inferoposterior OMI.
I sent the ECG to Dr. Meyers without any other information. He predicted RCA occlusion and tested the AI which he and Dr. Smith are training to identify OMI:
The patient did indeed have a 100% RCA occlusion, with peak troponin I over 10,000 ng/L. So under the OMI paradigm, expert-trained AI would help providers identify OMI and this patient would have had immediate cath lab activation, without any need to wait for repeat ECGs or troponin levels. Now let’s compare this with the existing paradigm to identify multiple preventable delays to reperfusion, which can be improved through the paradigm shift from STEMI to OMI.
1. ECGs labeled ‘normal’
Initially the patient had been on the medical wards for cancer treatment, and had been noted to have on/off chest pain with a troponin I that was mildly elevated around the 300 ng/L range. The day prior to the ECG above, the initial ECG below (ECG #1) showed minimal inferior T wave inversion:
(Marqette 12SL algorithm)
So the patient was being treated for "Non-STEMI" with anti-platelets and heparin, and was getting troponin levels done every 6 hours. The next day, the patient developed recurring chest pain accompanied by nausea, and an ECG was done (ECG #2, the same ECG from the top of this post, reproduced here to compare with prior):
Compared with the prior ECG, the ECG with active chest pain now has inferior T waves that are upright (pseudonormalization) and hyperacute, and the reciprocal change in aVL and ST depression in V2 are also new. This confirms infero-posterior OMI with even more certainty.
Many people want to believe that ECGs labeled ‘normal’ by traditional computer interpretation are unlikely to have clinical significance, and sometimes use the cardiologist's final interpretation (rather than patient outcome) as the reference standard.
But this is one of many cases highlighting this dangerous assumption. This ECG was diagnostic of OMI but was STEMI negative, i.e. STEMI(-)OMI, and was labeled and confirmed by a cardiologist as ‘normal’, so the patient was given nitro and reassessed.
Smith comment: Cardiologists, like all health care professionals who interpret ECGs, are a very heterogeneous group when it comes to ECG interpretation. Some are superb, some not so much. Cardiologists are indeed very smart and experienced and have a huge knowledge base, and do their best for their patients. But cardiology is a massive specialty, and ECG interpretation for OMI is a tiny part of it. It is so small that most cardiologists have never heard of "OMI," even though it is now in their literature (see the OMI literature timeline here). Reading the ECG for OMI is mostly pattern recognition. There are no measurements that are reliable. People who have good pattern recognition skills, know about OMI, and have devoted much time to learning the patterns in patients with known outcomes, are good at this skill whether they are ECG techs such as Vince DiGiulio (who later became an EMT) or whether they are the rare interventionalists, such as Emre Aslanger, who are aware of the many subtle ECG signs of acute OMI.
Therefore, cardiologists in general should not be used as the reference standard for interpreting OMI on the ECG, either in clinical practice or in research papers.
2. Serial ECGs
An hour later, with chest pain persisting, the ECG was repeated (ECG #3) along with a troponin:
This ECG still doesn’t meet STEMI criteria, and now interpreted as “nonspecific” by the computer and over-read, but it is still very specific for inferoposterior OMI. In the STEMI paradigm, patients with ischemic symptoms and ECGs that don’t meet STEMI criteria get serial ECGs. But STEMI criteria has poor sensitivity for acute coronary occlusion. As a result, patients with STEMI(-)OMI can get serial ECGs that are repeatedly diagnostic of OMI but repeatedly fail to meet STEMI criteria.
Often these ECGs NEVER evolve to meet STEMI criteria; ALWAYS they evolve too late for optimal timing of intervention. This leads to preventable delays to reperfusion: the patient waited another hour for the troponin results.
3. Initial troponin
But the troponin level decreased from 357 to 179 ng/L, because the previous mild troponin level was falling at the time of occlusion, and the repeat troponin was measured only one hour after symptom onset.
Troponin is like a rear-view mirror, showing damage from the past. Like the initial hemoglobin level in acute hemorrhage, the initial troponin in acute coronary occlusion can be falsely reassuring. This often presents as patients with acute coronary occlusion whose first troponin level is in the normal range, but in this case the troponin was previously mildly elevated and falling at the time of occlusion.
Smith: the definition of MI is based on clinical criteria and troponin. The troponin element is at least one troponin above the 99th percentile URL with rise and/or fall. So the patient can be diagnosed with acute MI and has recurrent chest pain. Those two elements alone are enough for cath lab activation even if the ECG does not show OMI (or if it does and you do not see it - or the computer does not see it)!!
According to the STEMI paradigm, a patient with serial ECGs labeled ‘normal’ and ‘nonspecific’ does not need reperfusion, especially with a falling troponin. But according to the OMI paradigm the patient had diagnostic ECGs and refractory chest pain, two indications for emergent reperfusion despite early and unreliable troponin.
Despite the troponin level the patient had refractory pain so another ECG was repeated (ECG #4):
Diagnostic of infero-posterior OMI for the third time, and now meets STEMI criteria for the first time two hours later. So the cath lab was activated and found a 100% RCA occlusion, as predicted from the first ECG. Troponin rose above 10,000 ng/L after the angiogram but was not followed to peak.
Below is a look at the 22 hour period surrounding the acute coronary occlusion, showing the troponin lag compared with symptoms and ECGs:
The
time prior to the occlusion included ECG #1 and serial troponins
from 284 to 269 to 357. When the patient developed acute coronary
occlusion they serial ECG #2-4, the last of which led to cath lab activation despite a falling troponin--and the repeat troponin after reperfusion was greater than 10,000.
Post-angiogram ECG (ECG #5) showed inferior Q waves, with T waves deflating and starting to invert, and resolution of reciprocal change in aVL:
ECG a few days later showed inferior reperfusion T wave inversion:
Take home
1. ECGs labeled as ‘normal’ by conventional machines (and confirmed by over-reads) can still be diagnostic of OMI, and identified by OMI experts and the AI they train
2. Serial ECGs looking for STEMI criteria can continually miss STEMI(-)OMI, leading to preventable delays to reperfusion
3. Troponin levels lag behind, and can be falsely reassuring early after occlusion
===================================
MY Comment, by KEN GRAUER, MD (4/21/2023):
===================================
I would add the following Take-Home Point to the excellent discussion of today's case by Dr. McLaren:
- PEARL: Still all-too-often ignored is the potential benefit derived from correlating the presence and relative severity of CP (Chest Pain) — to the timing of each serial ECG. It's EASY to establish that this invaluable information remains so often ignored — simply by the absence of notation either on each serial ECG or in the medical record, of whether the patient was (or was not) having CP (and if so — how severe?) — at the time each ECG is recorded.
-USE.png) |
| Figure-1: Comparison between the first 2 ECGs that were recorded in today's case. The TOP tracing ( = ECG #2) — is 1st first ECG that was shown above by Dr. McLaren. The BOTTOM tracing ( = ECG #1) — was obtained the day before ECG #2. The patient's CP was "on-off" at the time ECG #1 was recorded (See text). |
- Today's case began with us first being shown ECG #2. Even in the absence of clinical information — this tracing has to be viewed as highly suspicious of an acute ongoing event (exactly as it was immediately interpreted by Dr. Meyers and the Queen of Hearts AI program). There is simply no way that the flattened (if not, slightly downsloping) ST segment in lead V2 is ever "normal" (ie, The ST segment in leads V2,V3 should be gently upsloping — and often slightly elevated as a normal finding).
- As shown in the first illustration of Dr. McLaren's discussion — the computer interpretation of ECG #2 was, "normal ECG". Prior to development of the Queen of Hearts AI program — I had never seen a computerized ECG program sensitive in picking up subtle non-STEMI OMIs, such that: i) Clinicians always need to overread computer interpretations in context with the clinical situation; and, ii) Given that the patient in today's case was having severe CP at the time ECG #2 was recorded — definitive diagnosis of an acute infero-postero OMI (as per Dr. McLaren) has to be made!
- BUT — this patient had been admitted to the general medical ward the day before, at which time ECG #1 had been recorded. The reason for this patient's admission to the hospital the day before was for cancer treatment. His symptoms at that time were "on-off" CP — and he had some troponin elevation in association with ECG #1.
- NOW — Imagine today's case had started with us being shown ECG #1, with the knowledge that the patient was having "on-off" CP — and that an initial troponin value already showed some elevation. Even though there is no ST elevation in ECG #1 — the combination of intermittent CP (reduced at the time ECG #1 was recorded) — in association with shallow T wave inversion in all 3 inferior leads — should at least raise the possibility of spontaneous reperfusion — IF the treating clinicians were correlating the history with these subtle-but-real findings in ECG #1.
- To Emphasize: Even though ECG #1 is not diagnostic by itself of acute OMI — IF the treatment team was correlating ECG findings of this tracing with the history of intermittent CP and, with the fact that there already was an increase in troponin — Then closer observation would clearly be warranted (probably not still on a general medical ward) with repetition of at least several ECGs to ensure no active ongoing process.
- All that should need to be done is — i) To take into account that the reason ECG #2 was obtained, is that the patient was now having increased CP; and, ii) To compare lead-by-lead the appearance of ECG #2 with ECG #1 (as should be easy to do when you put these 2 tracings together — as shown in Figure-1). As per Dr. McLaren — comparison of these 2 tracings confirms acute dynamic ECG changes — which IF correlated to the history of now severe CP — should mandate prompt cath.
- The pathophysiology of acute OMI evolution often includes a period of spontaneous reperfusion, that sometimes is only transient before spontaneous reocclusion occurs. IF clinicians carefully correlate the presence (and relative severity) of CP with the timing of each serial ECG — they can usually figure out when spontaneous reperfusion has occurred because: i) This is most often accompanied by reduction (if not complete resolution) of CP; and, ii) Acute ECG changes (ie, ST elevation and depression) decrease, if not normalize on the way to developing the typical pattern of reperfusion T waves (ie, T wave inversion in leads that previously showed ST elevation). All the treating clinician(s) need to do — is correlate the presence (and relative severity) of CP symptoms with serial ECGs. Doing so often renders the diagnosis of acute OMI obvious.
No comments:
Post a Comment
DEAR READER: I have loved receiving your comments, but I am no longer able to moderate them. Since the vast majority are SPAM, I need to moderate them all. Therefore, comments will rarely be published any more. So Sorry.