Listen to the Clinical Chemistry Podcast
F.S. Apple, R. Ler and M.M. Murakami. Determination of 19 Cardiac Troponin I and T Assay 99th Percentile Values from a Common Presumably Healthy Population. Clin Chem 2012;58:1574-81.
Dr. Fred Apple is Medical Director of Clinical Laboratories at Hennepin County Medical Center in Minneapolis and Professor of Laboratory Medicine in Pathology at the University of Minnesota, School of Medicine.
This is a podcast from Clinical Chemistry, sponsored by the Department of Laboratory Medicine at Boston Children’s Hospital. I am Bob Barrett.
When it comes to measuring cardiac troponins, the 99th percentile is now firmly established as an essential component of the definition of myocardial infarction. However, there is no consensus on how to define a reference population or the total number of subjects needed to best determine that 99th percentile value. In the November 2012 issue of Clinical Chemistry, Dr. Fred Apple and his colleagues at the Hennepin County Medical Center in Minneapolis examined 19 different assays of cardiac troponins to help address this issue. Dr. Apple is our guest in this podcast.
Doctor, please remind us of the importance of the 99th percentile value of cardiac troponin I or T assays and the diagnosis of acute myocardial infarction?
The term acute myocardial infarction is used when there’s evidence of myocardial necrosis in the clinical setting of a patient who presents with an ischemic presentation. The cornerstone of the third Universal Definition of Myocardial Infarction, which is just published in September of this year, looks at the detection of a rise or a fall of cardiac biomarkers, specifically and preferably, cardiac troponin I and T.
To make the diagnosis, you need to observe at least one value of a rising value that occurs increased rising above the 99th percentile of a referenced population, determined, called a reference value.
What is so unique about this 99th percentile study compared to other reference value studies?
If you review the literature for the many, many cardiac troponin assays that are in the marketplace today, the majority of the studies are based on determining a 99th percentile based on a single population. So assay A might have a 99th percentile based on their population; a different assay might use a different population.
This study was unique and it was a very fun study, because I’ve been working on this for several years. The concept that we collected over 500 individuals and we used the same sample for all of the assays that we compared. In this case, we used 19 different troponin assays from multiple manufacturers in multiple formats: in the prototype high- sensitivity formats; in the contemporary sensitive assays we currently use in the marketplace; and in the points-of-care setting.
So, we had multiple assays in the same healthy, presumably healthy, population, which we’re able to do a direct comparison, assay to assay, using the same population.
In your Clinical Chemistry paper, what criteria did you use to define normality for your subjects? How do you know that they were really representative of normal individuals?
One of the critiques of the paper and I’ll self-admit it as a limitation: it’s very difficult to define a clear and clean normal reference population. What we’ve done for years, myself and other investigators, is you identify it often after an informed consent procedure. You identify the individuals who you give a questionnaire, a health questionnaire by interview, to find out, first of all, what their underlying pathologies are: Have they ever had a known cardiac- related disease? Do they have hypertension? Do they have renal disease? Do they have diabetes?
So you query the individual to the best of your knowledge or the nurses of that enrolled patients to find out if they have any underlying disease that might be directly or secondarily related to an underlying cardiac disease. If those individuals do come up with some type of underlying disease, you eliminate them from the study.
But what people have proposed, and this is a challenge to determine what a normal reference for the study is, is maybe we should start using surrogate markers such that, maybe we should use a hemoglobin A1c to really look for undiagnosed diabetes. Maybe we should use estimated GFR to look for renal insufficiency. Maybe we should take a blood pressure. Maybe we should do things like that, biomarkers such as an NT-proBNP or BNP natriuretic peptide marker to see if there’s underlying structural myocardial damage.
I think this would’ve added to the study if we had some surrogate biomarkers. It’s really cost-prohibitive to do what I’ll call the “Cadillac” type of approach to normal patients, which would involve really invasive testing, maybe echocardiogram or imaging and things like that. Those are really cost-prohibitive. So that was, I would say, the way we approach to identify a presumably healthy population we used in this normal range study.
The term high-sensitivity troponin assay has now been around for a couple of years and almost seems to be used indiscriminately. Can you tell us what makes an assay for cardiac troponin truly a high-sensitivity method?
The term high-sensitivity troponin assay is still a bit of a controversial term because it hasn’t been an evidence-based medicine term, but it’s what’s been adapted by the experts both in laboratory medicine and in cardiology. When we refer to high-sensitivity, the concept that people have learned to kind of move forward towards in that concept is two things: one, is that we’re going to be able to use an assay that will enable to measure troponin values in greater than 50% of a presumably normal population, such that if we enroll a thousand patients, we’re going to be able to use this assay and the number we generate will be greater than the limit of detection of the assay in at least 50% of the individuals.
The second criteria is that the imprecision of the assay would be at 10% at less at the 99th percentile. So if you look at the paper we published in Clinical Chemistry and you look at all the high-sensitivity assays, whether they’re prototype or the Roche assay that’s used in Europe, the rest were the prototype assays for high-sensitive troponin I, which are not FDA cleared or available in Europe yet, you’ll see that they all measured at least 80% of the normals and all of them had an imprecision at the 99th percentile, 10% or better. So therefore, it defined a high-sensitivity assay.
Now, you examined 19 different methods for measuring cardiac troponins, including some point-of-care procedures. What similarities and differences did you observe?
If we take a look at this paper, especially if we look at Figure 1 from this paper, it’s a plot to y-y x-axis plot, which on one of the axis, we actually look at the 99th percentile concentration, and the other axis, we look at the percent measurable concentration, and we divide the figure up into three zones: a point-of-care zone, a contemporary-sensitive assay, which are assays we use in the current practice, and then a high-sensitivity assay zone. And the key thing you see, first of all, with the troponin 99th percentile is that there is no harmonization or uniformity or any standardization across assays, and these are represented by the circles which show you that every assay, almost 100% of the time, has a different 99th percentile value.
That’s a very important concept to understand because currently, even though there’s work by several international committees to try to standardize assays, it points out the importance of that if you’re going to use an assay, you have to use the 99th percentile based on that individual assay and not try to compare it and use a similar 99th percentile for a different assay.
And this is an important thing to consider, because more and more hospitals are bringing up point-of-care assays into, let’s say, an emergency medicine situation. So if your 99th percentile for your point-of-care emergency medicine assay differs than your central lab assay, you don’t want to be reporting those out in the same medical center, because physicians are going to get confused. They don’t care about assay A and B, they just want to know if it’s positive or negative, and if you start comparing numbers, that’s going to get mixed-up. That’s number one.
The second thing that was quite evident from this figure and from this paper is that the large majority of our current contemporary assays and the point-of-care assays measure less than 10% of the normal populations. Therefore, out of these 500 plus individuals we looked at, we barely saw anything that gave a measurable concentration above the limit of detection. However, as I mentioned earlier, when we go to the high-sensitivity assays, all of these assays were able to give us values that were greater than at least 80% measurable values in the normal range.
Well, I guess the question is, do we really need 19 different methods for troponin, each with a different 99th percentile cutoff?
Simple answer is, we don’t need them in the laboratory, but the manufacturers are able to develop their own assays and their own configurations. So, it’s a free world marketplace out there so where every system has to be understood by the laboratory what they’re using. So you have to know your assay.
Can you comment on the lack of agreement among the cardiac troponin I assays you studied? You found differences in assays, even from the same manufacturer.
Yes, and I think that is an important point to understand, that every instrument has a different configuration, and every instrument may design the same antibodies but might have one as a capture and one as a detection, and reagents are formulated differently depending on how these instruments analyze the specimens.
So, the point here, which is very important to understand, is that just because you have a company called company A and they might have three assays, it doesn’t mean that their numbers are always going to agree and that is a very important thing not to miss.
In your study, you show significant differences between genders for many of the assays studied. Why is that important?
The gender differences for 99th percentile in males and females are critically important for both diagnostic purposes and risk purposes. If you look at Table 1 in the study, it shows across all the high-sensitivity assays, for example, a 1.2-4.0-fold difference between males and females.
Also, if you look at some of the sensitive assays in 81 point- of-care assays, there’s a gender difference. This is important, because based on the universal definition, if you’re looking at a rising pattern and trying to make a decision on a concentration of troponin that rises above that 99th percentile, if you had a female patient, if the initial value is 12, and then the second value six hours later rose to 20, if you’re using this total concentration, that would have never been an increase. If you look at the female concentration, which I said was 15, it would’ve met the definition for the universal definition.
In addition, when patients are looked at very carefully for risk stratification, you might look at it by quartiles or by just a single cutoff for risk. If you’re using again 15 female and 36 male and you’re not looking at males and females separately, again, the risk stratification potential would be dramatically different. You could see 20% to 30% differences, because where the assays and especially the high-sensitivity assays are moving, we’re now finding the ability, just like we found for CRP, is we’re able to discern risk even within the normal reference population. So if you don’t differentiate men from women, you will get overlap and miss that great potential to risk stratify.
Well, finally doctor, what roles should high-sensitivity cardiac troponin assays have in clinical practice for diagnostic accuracy, risk assessment, and potential use in primary prevention?
Well, I think I address the powerful role it’ll have in diagnosis, being able to measure values at lower concentrations. We are starting to see a trend that we’re going to detect rising patterns earlier. Where we used to wait up to six hours, now we’re seeing a diagnosis in both ruling in and ruling out, a window of two hours to three hours after the presentation.
Second, we talked about risk stratification of patients who have acute coronary syndrome and other cardiac-related pathologies. We’re doing a better job risk stratifying even within the upper normal reference range and those patients slightly above the 99th percentile, but most promising, there’s new literature coming out for primary prevention.
Typically in primary prevention, most people use a Framingham risk score, which we’re trying to figure out, could we add biomarkers that’ll optimize risk assessment of patients who are in the mid- to intermediate range of risk? So, there’s a great search now for looking at adding a biomarker to improve risk stratification. This, we’re talking about, in normal individuals. And there are several studies out there that have shown that even in healthy populations, high-sensitivity troponin I and T improve risk stratifications for cardiovascular disease related to acute coronary syndrome, as well as for heart failure, improving risk stratification over a 5-10 year period in a 4.0-6.0-fold higher relative risk. So one of the thoughts here is, is it possible with these high-sensitivity assays to potentially compare it and add it to maybe what the Framingham risk score is based on, which is other biomarkers like cholesterol, and clinical features like blood pressure, diabetes?
So I think the potential here is that these high-sensitivity assays of troponin I and T may make an excellent prediction and additive value to what we currently use in risk stratification. In my view is in the future, this biomarker, using a high sensitivity assay, may be something we baseline our young adults and young children on, because I think we’re really close to getting to the final rendition of these assays. Once we can measure everyone’s value with precision, we’ll be able to track changes very precisely over time.
Dr. Fred Apple is Medical Director of Clinical Laboratories at Hennepin County Medical Center in Minneapolis and Professor of Laboratory Medicine in Pathology at the University of Minnesota, School of Medicine. He has been our guest in this podcast from Clinical Chemistry. I’m Bob Barrett. Thanks for listening.