The most sensitive test for diagnosis of AMI/ACS  is probably Troponin I . The 3 important features of ACS are chest pain , ecg changes & cardiac biomarkers . ECG changes may take time & among cardiac biomarkers including CPK-MB , Trop T & I , Trop I is probably the best .

Hi John, from my laboratory point of view, every clinical laboratory has to set its own cut off to discriminate patints. When I worked in a cardiology hospital, I had bad experience when we changed “second generation” troponins into “high sensitivity” ones. After several months clinicians get used to new method, but firstly we had to set empirically the cut-off for hs-troponins.

Of course, a clinician must be aware that 20% elevation above the cut-off is always a decision making point. What is suggested in such case in symptomatic patients? To perform the second analysis to confirm troponin elevation within next 3 hour. What is recommended in AHA and ECS guidelines. What I can recommend, I do it to my students, to use the second marker. Which one?

Those which are proceeding to troponins: copeptin (CT-proAVP a peptide derived from a pre-pre-vasopresin peptide) or h-FABP (heart specific fatty acid binding protein).

I hope that it has helped.

Czesc Ewa i dziekuje (I spent 4 years living in Kielce in the 90s).

I'm interested you set your own cut-off.  Are you using hsTnT or hsTnI?  How do you set your own cut-off - is it based on achieving a maximum False Negative rate for AMI (or MACE) in your hospitals, or is it a particular percentile from a healthy population?

I'm involved in research to enable early discharge based on a 0 & 2h negative hsTnI or T, a negative ECG, and negative risk score (see Than, M. P.,  et al. (2014). A 2-hour diagnostic protocol for possible cardiac chest pain in the emergency department: a randomized clinical trial. JAMA Internal Medicine, 174(1), 51–58. doi:10.1001/jamainternmed.2013.11362).  2 hours is too short to see a definite rise or fall in troponin to rule-in AMI.  So in those with two elevated troponins, a third at ~6+ hours is probably needed.

We are also looking at Copeptin and hFABP.  I'm interested in the comparative time courses of these compared to hsTnI or hsTnT.  Do you know of any work measuring these at multiple time points?

This is not an easy question to answer....If one wants to experience the nirvana of a 0% FN rate (essentially shooting for 100% sensitivity), then be prepared to accept an increase in the FP rate resulting in more patient testing and admissions that would not have corresponding yield.  I also would suggest that the "acceptance level" for FP rate will vary based on the physician specialty.  For instance, an emergency physician will no doubt want to have a very high negative predictive value. This translates to the emergency physician's willingness to sacrifice the expected increase in low yield admissions and testing in order to decrease false negative discharges from the ER resulting in high morbidity/mortality and medico-legal exposure.  I would suggest that a diagnostic test (or compilation of tests) with specificity of approximately 95% with a sensitivity not dropping below approximately 85% might be acceptable. I would be looking for a 90-90 approximate balance.  Anyone have thoughts on this????

I agree totally with David. It is always a dilemma to how to achieve 100% sensitivity without falce positive results.

Using one marker it is never possible, there is no ideal cardiac marker for AMI, because troponins are makers of cardiomyocyte damage, not myocardial ischemia (troponins are cytoskeletal protoeins which are release from completely damaged cell, within several dozen minutes after occlusion). Thus, I can propose a multimarker approach. Clinicians usually do not like such approach, because this approach is cost intensive.

How to set a cut-off for troponins? Or for any single marker you want to apply for disease discrimination. Thar is a 95-percentil or 99-percentil method or, what I prefer most, the Receiver Operating Curve (ROC) analysis.

The first one does not apply patients. You can set you our cut-off using a representative (with sex, age, and race in some cases) group of apparently healthy subjects. Then more than better. The measured parameter is analyzed to have the value of 95- or 99-percentile, to beset as cut-off. Every value above this cut-off discriminate “healthy” from “not-healthy” subjects. Please notice, “not-healthy”, but not “AMI”, because AMI group was not a subject of analysis.

The second method needs a representative “healthy” group which is matched with respect of age and sex to an AMI group. The “marker value” must be measures in both groups and lined up from the highes to lowest in relation to AMI-presence. The probability of false positives and negatives must be calculated using ROC formula and the cut off value can be set with specificity of approximately 95% with a sensitivity not dropping below approximately 85%.

I can suggest you some papers where ROC analysis is successfully used, both for new and old cardiac markers.

Please see other papers:

http://www.ncbi.nlm.nih.gov/pubmed/19555842

Hi John! I think you know the answer!!

It's interesting that the conversation has jumped to troponin levels but perhaps that is the thread. Your question is though, quite generic - I imagine that was your intention.

Interestingly, I think it depends who you ask. As you know Martin's survey of many emergency physicians suggested that the majority accepted a 'miss rate' of somewhere between 0.5 and 1.0% (at least for short term MACE). When I've done a 'straw poll' of cardiologists, however, I tend to get a slightly lower number. I think that ED physicians are a bit more pragmatic!

I once gave a talk to a group of aspiring hospital administrators and they nearly all suggested that the miss rate should be 0%!!!

I've been in clinical practice for long enough to know that nothing works 100% of the time.

We've been thinking about this quite a lot in the context of identifying a very low risk (but adequately sized) cohort amongst those presenting with undifferentiated chest pain. What we've come up with is a rate of 0.3% (for a 'hard' MACE endpoint) with a non-inferiority margin of 0.5%.

I don't have space here to explain how we came to this conclusion!

Hi Will

Yes, I know the survey's answer (For everyone else, see: Than, M. P., , et al. (2013). What is an acceptable risk of major adverse cardiac event in chest pain patients soon after discharge from the Emergency Department?: a clinical survey. International Journal of Cardiology, 166(3), 752–754. doi:10.1016/j.ijcard.2012.09.171).  Part of my purpose was the question was to reach out to find if there was other thinking out there.

I will enjoy a discussion about how you came with 0.3%. One of the difficulties I'm finding in this field is that compared to anything else I've done the sensitivities are so very high that improving on them is v difficult and requires very large numbers.  

I agree. What we need now is sensible dialogue around how this stuff is translated pragmatically into practice. In 2000 Pope published missed ACS rate of ~4% http://www.ncbi.nlm.nih.gov/pubmed/10770981?dopt=Citation

Interestingly, in the discussion that was considered acceptable (in the face of other papers that had suggested that rates might be much higher) but now that paper is usually cited to demonstrate how poor chest pain assessment was at that time - we've come a long way in 15 years and sometimes I think the pace of that has given us unreasonable expectations of what is, after all, a complex diagnostic process.

If we translated this to RCTs of advances in treatment for a disease and we had shown an absolute reduction in mortality or morbidity from ~4% to

Hi everybody!

Excellent question, outstanding answers and comments!

I posed this same question to my ex-associates when introducing the first Chest Pain Unit in Brazil in 1996, and then at the Abstract Sessions of the American College of Emergency Physicians congress in 1999, again at the European Society of Cardiology congress in 1999 and 2000, and again at the American College of Cardiology congress in 2002. At that time we only used CKMB as a necrosis marker.

As we can see almost 2 decades later, and approprietely mentioned by all of you, there is no definite answer for this practical and very important question.

The introduction of high-sensitive troponins brought this questioning again but made things even more difficult (not to say, worse...) in the decision-making process in the ER.

I cannot agree more with Will's last paragraph but how to achieve that is the one-million dollar answer!