March 22nd, 2011

Study Explores Lowering the Troponin Diagnostic Threshold

Lowering the troponin diagnostic threshold can significantly improve outcomes after MI, claim Scottish investigators in a report published in JAMA. More than 2,000 patients with suspected ACS were studied. Sixty-four percent of patients had troponin concentrations below 0.05 ng/mL, 8% had concentrations from 0.05 to 0.19 ng/mL, and 28% had concentrations 0.20 ng/mL or higher. By lowering the diagnostic threshold from 0.20 ng/mL to 0.05 ng/mL, the risk of death and recurrent MI was reduced from 39% to 21% in the group with troponin concentrations from 0.05 to 0.19 ng/mL during the validation phase of the study.

The authors write that “the appropriateness of continuing to lower the threshold of plasma troponin assay concentration to define increasing numbers of patients with MI may be questioned. This concern relates to the potential to reduce specificity and increase false-positive diagnoses of MI. Our study supports the contention that this is not the case, rather the concern relates to the potential for misclassification of high-risk patients through the use of outdated diagnostic thresholds.”

8 Responses to “Study Explores Lowering the Troponin Diagnostic Threshold”

  1. This is an interesting study design, which capitalized on a rigorous evaluation of the implementation of a new troponin assay. However, I have several concerns about the interpretation of the study by the authors.
    1. The study really tests differences between an obsolete troponin assay and a contemporary one. It does not evaluate newer highly sensitive troponin assays that have recently been the focus of a number of research studies. It seems obvious that an assay as insensitive as the one the authors were previously using would perform poorly and misclassify individuals with actual MIs. However, it is less clear that continued movement towards improved sensitivity will improve ER care. I am concerned that implementation of highly sensitive troponin assays will produce many more false positive MI diagnoses than real MIs detected.
    2. In the U.S. the typical “rule out MI” patient that is admitted to a low risk observation unit has a probability of “ruling in” for MI is well below 5%. This contrasts sharply with the MI rate in this study of > 30%. In a low probability situation, the balance of false positive vs true positive diagnoses will be unfavorable.
    3.I think it is implausible that improvement in outcomes would occur merely with implementation of the lowered troponin cutpoint. The improved outcomes after implementation of the more sensitive assay were out of proportion to fairly modest differences in treatments received after the change. For example, although administration of dual antiplatelet therapy differed, revascularization rates did not seem to change, despite the much higher rate of referral to a cardiologist. Moreover, the numbers of patients and events were far too small to draw conclusions from.

    In the end, it is certainly reasonable to upgrade to contemporary troponin assays and to define MI at the 99th percentile cutpoint from a normal population. However, whether highly sensitive assays will improve care or cause confusion remains to be seen. The more sensitive the assay, the more important the role of clinical judgement in determining whether the troponin elevation is from acute ischemia or another cause.

  2. Saurav Chatterjee, MD says:

    Advocating clinical judgement is of course pragmatic and essential-however once there is a consensus to define a cutoff value,no matter how unlikely a candidate is for having ACS-if his cardiac enzymes values exceed the cut-off,I doubt many physicians will agree to have the patient return home without repeated testing for confirmation/longer duration of in-hospital stay……….

  3. David Powell , md, facc says:

    Quirky stuff. Must have been a high pre-test likelihood. Event rates indeed staggering. Also, the risk gradient as troponin increases seems fairly flat: a 21% hard risk for trop between 0.05 and 0.2 and a 24% risk for > 0.2 (much higher on average…one would think).
    As long as the population- based cut- offs remain the same, there should be no decrease in specificity for MI for the more sensitive assays. Theoretically, its the same ROC curve and same cutoff. This should be increasingly the case as detection threshold has decreased and reliability increased (resulting in more uniform population-based threshholds).
    So the only point of more sensitive assays now relates to other potential uses. These include general population screening, CHF/cardiomyopathy risk stratification, and early detection of chemo-related cardiotoxicity ( where changes within the sub-MI zone are important).

  4. I see the biggest challenge in normal value reference limits. First we have to be sure what normal values are, only then can we make some conclusions. What about the correct definition of cutoff value then? Lowering these ill-defined cut-off values for troponin will lead to positive results in patients who do NOT have ACS, but have chronic troponin elevation or the elevation due to other causes than ischemia. Or even laboratory interference may drive the results to either normal or pathologic values.
    Therefore I would like to cite Dr. Allan Jaffe from his Clinical Chemistry Podcast on High-Sensitivity Troponin Assays.
    1/ … presently most of the assays do not define values for most of the normal population.
    2/ As we begin to use these new high-sensitivity assays, there may come a time when we are detecting such small amounts of injury that that may longer be the case. So, we have got to go back and do the clinical studies, not just of these patients who present acutely, but about patients who might have more chronic elevations as well.

  5. Robin Motz, M.D., Ph.D. says:

    This is a restatement of the old argument: is it more dangerous for the patient to have a false positive or a false negative test? The answer determines where we set the cutoff. Since heart attacks can be fatal or lead to permanent cardiac dysfunction, I would prefer to lower the threshold so as to have fewer false negatives at the expense of having more false positives. Which would you prefer if you were the patient?

  6. David Powell , md, facc says:

    Agree…the normal population range for the tests must be robustly documented. The key info is positive and neg predictive values, and this will depend on pretest likelihood. This is where the” universal MI” cutoffs can fail us since, since they do not take into account the pretest likelihood ( symptoms are a dichotamous variable). So if the story for ischemia is clinically poor, our MI cutoff should be higher… say 4x the population-base 99% level. But I do not think the current low and sensitive cutoffs for ER chest pain will be decreased regardless of the pre-test likelihood. The patients’ whose pretest likelihood is so high to even consider a lower cutoff would be obvious unstable angina cases. The more sensitive assays should not further change the current low cutoffs.
    I agree with the research part, as the troponin cutoffs seem too low after revascularization, as shown in a recent CMR study using LGE as a guide
    Further complexity arises if one considers the timing of troponin assessment relative to the ischemic symptoms. I believe our most recent jump in assay sensitivity increases the negative predictive value of undetectable troponin 4 -6 hrs after symptom onset. This may obviate repeat draws and facilitate early discharge.?

  7. Travs Harrell, MD says:

    Do we need “better” troponin assays? As a junior internist, I can’t say that I have seen many patients who have had a negative outcome from a “false negative” troponin test, or any for that matter. On the contrary, I have seen plenty of false positives for a variety of reasons (CKD, cardiomyopathy, PE, myocarditis, sepsis, lab error, etc.).

    Does anyone on this blog routinely see false negative troponin results in patients presenting with chest discomfort or dyspnea?

    Competing interests pertaining specifically to this post, comment, or both:

  8. It’s my impression that serum troponin is obtained in almost all patients presenting to the ED with medical or surgical issues, with an elevated level prompting hospital admission in order to “sort it out.” Last night, a serum troponin was obtained in an ED patient who presented for evaluation of epistaxis. It was elevated, which means I met her this morning in the CCU. “Troponin gone wild”