Rapid algorithms and early rule-out with novel Gen 6 high-sensitivity troponin assay

Guest speaker: Luca Koechlin, MD - Department of Cardiac Surgery, Allianz Herzchirurgie Zurich, Cardiovascular Research Institute Basel (CRIB)

*This transcript was generated using an AI-based transcription tool and may contain errors. It reflects the spoken content of a live webinar and has been lightly edited for clarity.

Ladies and gentlemen, it is my great pleasure to be here today and to present to you the first data, the first clinical data of the novel Gen 6 high-sensitivity cardiac troponin, and especially about the performance of the 0/1- and 0/2-hour algorithms.

Now, these are my disclosures. However, the most important disclosure slide is this one here. As mentioned before, I am very happy and grateful to be a part of the fantastic research team of the Cardiovascular Research Institute in Basel, under the direction of Professor Christian Mueller. And without all these people, research like the one I'm presenting today would not be possible.

Now, in today's talk, I want to first give a short introduction about high-sensitivity cardiac troponin in general in the early diagnosis of acute MI. And then, of course, focus on the performance of the novel Gen 6 high-sensitivity cardiac troponin and then conclude the findings.

Now, about 10% of all ED consultations are for suspected acute MI, and myocardial infarction still is a common and acute life-threatening condition. And of course, you all know that rapid diagnosis, according to the treatment scheme "time is muscle," is critical. But us ED physicians also do know that it's not only about the rapid diagnosis and the rapid rule-in, but also about the rapid rule-out of myocardial infarction, which is of major medical but also economic importance. Timely detection of other causes and alternative causes of chest pain are important, and it helps us to consider outpatient management in many of these patients.

And when we look at the numbers, then these are really extreme. We talk about 20 million patients per year presenting in Europe and the US with suspected myocardial infarction. And only a minority of these patients finally will be diagnosed with acute MI. And therefore, an early triage decision in the emergency department is critical, and we will see this later. According to the concept of the ESC 0/1- and 0/2-hour algorithms, this is a triage to rule-out, rule-in, or the so-called observe zone.

So, now what is known about cardiac troponin in general? We do know that cardiac troponin is more sensitive and a specific marker of cardiomyocyte injury compared to CK, CK-MB, and myoglobin. The introduction of high-sensitivity cardiac troponin has increased the diagnostic accuracy for MI at the time of presentation as compared with conventional assays, and this is important because it helps us to detect smaller myocardial infarctions, which may be missed with conventional assays. It helps us to detect myocardial infarction earlier, and this is especially important in the vulnerable subgroup of early presenters, so patients presenting within one, 2 or 3 hours after chest pain onset. And it allows us for a more rapid rule-in and rule-out of myocardial infarction. And therefore, we do have a very strong Class IB recommendation by the current ESC guidelines to measure cardiac troponin with a high-sensitivity cardiac troponin assay immediately after presentation in patients with suspected non-ST-elevation myocardial infarction.

Now, when we talk about rapid triage algorithms, the introduction of high-sensitivity cardiac troponin assays has allowed us to reduce the time interval to the second cardiac troponin assessment. And this could be shortened when we use high-sensitivity assays. This helps us to reduce the delay to diagnosis, translating this into a shorter stay in the emergency department and therefore lower costs. And also here, we do have a very strong recommendation by the current guidelines of the European Society of Cardiology to use the 0/1- or 0/2-hour algorithm to rule-out or rule-in patients with non-ST elevation myocardial infarction.

Now, obviously you are familiar with the concept of the 0/1- or 0/2-hour algorithm. But I want to mention it here again because it's quite easy. And this is, I think, what is very important in the busy emergency department. We can rule out patients for myocardial infarction if they have a very low troponin concentration at presentation or a low troponin concentration at presentation, in conjunction with a small delta between the two measurements. On the other hand, patients are ruled in for myocardial infarction if they have a very high troponin concentration at presentation or a relevant delta within the two measurements. All patients which are not triaged into rule-in or rule-out are triaged to the so-called observe zone. So, I'm not sure that "observe zone" is the right word for it, but we should further investigate these patients with serial troponin measurements and imaging.

Now, the optimal thresholds for rule-out were selected to allow for a minimal sensitivity and negative predictive value of 99%. Why? Because we want to have the rule-out approach as safe as possible. We do not want to rule out patients who finally have a myocardial infarction. On the other hand, the optimal thresholds for rule-in were selected to allow for a minimal positive predictive value of 70%.

These algorithms were developed in large derivation cohorts and then validated in large validation cohorts. And this is important: whenever you are using the 0/1- or 0/2-hour algorithm, you have to use assay-specific cutoffs. So it's really depending on which assay you are using in your institution. And this is an overview about all the at-the-moment available high-sensitivity cardiac troponin assays, including point-of-care assays where we do have assay-specific cutoffs for the 0/1- or 0/2-hour algorithm.

Now let's talk about the novel Gen 6 assay. The well-known high-sensitivity cardiac Troponin T Gen 5 assay was the first high-sensitivity cardiac troponin assay to be introduced in clinical practice, and is the most extensively validated among the different high-sensitivity troponin assays. It has a very high diagnostic accuracy for myocardial infarction already for measurements performed at presentation. It has therefore enabled the derivation and successful implementation of accelerated 0/1- and 0/2-hour pathways.

However, the Gen 5 assay also shows some limitations. We are talking about lower analytical sensitivity, lower precision in the low-to-normal range, and a higher proportion of patients with chronically elevated levels indicating myocardial injury. And you also do know this leads to uncertainty in the emergency department, and maybe we are doing unnecessary investigations in these patients. And furthermore, from an analytical point of view, interference with hemolysis is much higher with Gen 5 compared to other assays. So these limitations led to the development of the novel Gen 6 high-sensitivity cardiac troponin.

Now, as usual, when a novel assay is introduced into the market, it is mandatory that we analyze and assess its diagnostic performance in large, well-structured studies. And this is what we did in our study. We aimed to assess the analytical agreement between Gen 5 and Gen 6, to assess the diagnostic performance of Gen 5 and 6 in a direct comparison. So all the results you will see later, always for patients where we had measurements for Gen 5 and Gen 6. We aimed to derive and validate using optimized assay-specific cutoffs for the single rule-out strategy, but also 0/1- and 0/2-hour algorithms, and assessed risk stratification when using these algorithms.

Now, to address these questions, we used data from the well-known International Multicenter APACE trial, where patients are included presenting to the emergency department with chest discomfort, and defined a diagnosis well adjudicated by two independent cardiologists according to the Fourth Universal Definition of Myocardial Infarction and current guidelines. Furthermore, we performed an external validation of these findings in the cohort from TREDIT, present, and the BACC trial. And you will explain this in more detail.

Now, when we look at the patients included in our study, then you see here that we had more than 3300 patients included for the analytical analysis here. We also included patients with ST-elevation myocardial infarction. These patients were excluded for the diagnostic analysis because, as you know, troponin is not important for diagnosing ST-elevation myocardial infarction. And for the derivation and validation of the 0/1- and 0/2-hour algorithm, only patients were included where we had measurements at presentation and after one and two hours, respectively. You see that the cohort is a typical chest pain cohort with a median age of about 60 years. About 35% of the patients were women, and a similar amount of patients were early presenters, meaning they presented within three hours after chest pain onset.

Now, when we look at the density plot and just at the distribution of the concentrations in all these patients for Gen 5, then we see that about 50% of the patients had concentrations between the LoD and the upper reference limit of the 99th percentile, which is 14 nanograms per liter for Gen 5. There's a different picture. When we look at Gen 6, more than 60% of the patients had concentrations between the LoD and the reference limit, which is 27 for Gen 6, and this resulted in a significantly lower amount of patients with concentrations above the 99th percentile and therefore indicating myocardial injury. And I think this is a very important first message for you as clinicians. When you use Gen 6, the amount of patients with concentrations above the 99th percentile indicating myocardial injury is significantly lower compared to Gen 5. Again, these were the same patients' measurements of Gen 5 and Gen 6. So really a direct comparison of Gen 5 and Gen 6.

Now, what about the diagnostic accuracy for troponin at presentation for non-ST elevation myocardial infarction in all-comers? Very high area under the curve of 0.93 for Gen 6. Comparable performance for Gen 5 and Gen 6. And when we look at only the early presenters, again, patients presenting within three hours after chest pain onset: also similar performance of Gen 5 and Gen 6.

Now, what about the single-measurement rule-out strategy? And I do not want to go into too much detail because Fred here is addressing this topic later in more detail in his cohort. But when we look, and this is very important in all-comers, again, irrespective of chest pain onset, when we look in more than 3160 patients, when we use a cutoff below six nanograms per liter when using Gen 6, we can rule out 30% of our patients with a negative predictive value and a sensitivity of 100%, so you do not miss one myocardial infarction. When we compare this to Gen 5 and the validated cutoff for Gen 5, which is below five nanograms per liter, the amount of patients which can be ruled out is significantly lower with only 13%, while maintaining a negative predictive value and sensitivity of 100%. So I think a second very important message for you as clinicians.

Now let's talk about the 0/1-hour algorithm. Now you have seen the concept before. Here you see the specific cutoffs which were derived in the APACE cohort for the 0/1-hour algorithm. So patients can be ruled out for myocardial infarction if they have a troponin concentration below eight nanograms per liter at presentation, or below eight nanograms per liter in conjunction with a one-hour delta below two. On the other hand, patients are included or ruled in for myocardial infarction if they have a troponin concentration greater than or equal to 112 nanograms per liter or a one-hour delta over ten nanograms per liter. And applying these cutoffs, we can triage 56% of our patients into "ruled out" with a very high negative predictive value and a very high sensitivity for rule-out. So we only miss one patient with a myocardial infarction. On the other hand, about 20% of the patients can be ruled in with a very high positive predictive value and high specificity for the rule-in approach. About one quarter of the patients remained in the so-called observe zone, and these findings were confirmed in the internal validation. Again, very high safety for the rule-out approach, very high accuracy for the rule-in approach.

Now, what about the 0/2-hour algorithm? For some emergency departments, it's easier to use a 0/2-hour algorithm. And also here of course slightly adapted cutoffs. But again, very high negative predictive value and very high sensitivity for the rule-out approach, while very high positive predictive value and specificity for the rule-in approach, and also these findings were internally validated, but not only internally validated but also externally validated. And again, big thank you to the group from Edinburgh for externally validating our findings for the 0/1- and 0/2-hour algorithm. And again, you see very high safety for the rule-out approach, high accuracy for the rule-in approach for the 0/1- and 0/2-hour algorithms.

Now I want to mention that the ESC 0/1- and 0/2-hour algorithms are diagnostic tools for the triage of patients with non-ST elevation myocardial infarction. But I also do know that prognostic performance of these algorithms is always an important topic. You see, the five-year survival rate according to the triage of the patients, according to the 0/1-hour algorithm: five-year mortality in the rule-out group was very low with 2.7% and, as expected, significantly lower compared to patients in the observe zone and rule-in zone. I also want to highlight the quite high mortality in the observe zone. So once again, these are not patients which should just wait for hours in the emergency department while we are observing them. We really need further investigation tools for this vulnerable patient cohort.

Now of course, there are limitations of our study. We excluded patients in cardiogenic shock and with end-stage renal disease requiring dialysis. There is additional research needed in other settings than the emergency department, especially thinking about coronary care units or primary care settings. The study may have underestimated the possible superiority of Gen 6, as Gen 5 was part of the central adjudication and Gen 6 was not. And last but not least, prospective implementation studies incorporating these cutoffs into clinical decision-making are warranted.

But do you? Ladies and gentlemen, I think we can. And I would like to conclude that using the novel high-sensitivity cardiac Troponin T Gen 6 assay in the Early Diagnosis of Myocardial Infarction, fewer patients had Gen 6 concentrations above the upper reference limit compared to Gen 5. Diagnostic discrimination for Gen 5 and Gen 6 was very high and comparable. A single rule-out strategy allows you to rule out 30% of your patients with 100% sensitivity and negative predictive value, irrespective of chest pain onset. So including early and the so-called very early presenters. And last but not least, assay-specific cutoffs for the 0/1- and 0/2-hour algorithms are available, and therefore early triage on the emergency department, using these from the guidelines, recommended algorithms is possible, and these findings were consistent in several subgroup analyses as well as internal and external validation. And with this slide, I want to conclude and thank you very much for your attention.

Guest speaker: Dr Freddy Thurston -  Clinical Research Fellow in Cardiology, Centre for Cardiovascular Science, University of Edinburgh, UK

*This transcript was generated using an AI-based transcription tool and may contain errors. It reflects the spoken content of a live webinar and has been lightly edited for clarity.

My name is Freddy Thurston. I'm a PhD student and clinical research fellow at the University of Edinburgh, as mentioned by Professor Mills. And I'm very grateful for the opportunity to talk to you today about our work using the new sixth generation high-sensitivity troponin T assay for ruling out heart attacks early on in a patient's ED journey. So these are my disclosures.

And so, before I joined the dark side, as it were, and started doing a PhD in cardiology, I worked for two years as a junior doctor in the emergency department in Edinburgh. And so I am sure that I do not need to tell this crowd how big a problem emergency department crowding can be. We've seen, particularly in the UK since the Covid pandemic, a significant rise in the number of patients waiting more than 12 hours. Emergency department overcrowding is highly demoralizing to staff and patients, often attracting significant negative attention in the media, but most importantly, we now have data that shows the longer you wait in the emergency department, the higher your risk of death at 30 days is.

So how do we tackle this? And one of the models that's been suggested to conceptualize how we might solve some of the problems that drive ED crowding is the Input-Throughput-Output model. So obviously, input and output are pretty intuitive. Throughput refers to the amount of time it takes to complete the diagnostic assessment of the patient, including processes such as history taking, assessment, blood tests, and so on. High input, low output, and slow throughput all contribute to rising numbers of patients in the department. And in my experience, and I'm sure many of you have had the same experience, times when occupancy of the department significantly exceeds capacity, throughput becomes very difficult. You know, you can't find a space to see new patients, you can't find space to get the tests done, it takes longer to do everything that needs doing for any one patient. And that really creates a vicious cycle in which flow can really grind to a halt.

We think that the assessment of patients with possible MI offers a particular opportunity to streamline throughput. As Luca showed you in a very nice visualization, there are a lot of patients who attend the ED with possible MI, suspected cases. Fewer than 1 in 10 of them actually have a myocardial infarction. Naturally, it's a serious diagnosis. The concern around the consequences of missed MI means that conventionally, we've always done serial troponin tests to make sure we weren't missing any kind of rise in these patients.

So troponin tests have obviously been improving with time, culminating with the recent sort of release of the sixth generation high-sensitivity assay from Roche. And what we've seen is that these assays have become more and more accurate at very low concentrations. And for the sixth generation assay, this limit of quantitation of the assay, the concentration you can measure accurately with less than 10% coefficient of variation, is now as low as three nanograms per liter. And what I hope this graph kind of conveys, the idea is that as this analytical precision increases in the very low range of concentrations, the early risers you see following an MI can be detected more reliably and at a shorter time interval following the event. And we think this is where the real value of high-sensitivity troponin assays over the old contemporary assays has come. It is that this sensitivity for MI at earlier points in time means that we can rule the diagnosis out with high confidence earlier on in the patient's journey.

So, I'm very lucky to work in a brilliant research group that's previously derived and validated this pathway, known as the High-STEACS pathway. And this was designed to leverage the properties of high-sensitivity troponin in order to triage patients with possible MI. So, patients come in, they have troponin and ECG measured at presentation. And the troponin concentration is used to categorize them as either low, high, or intermediate risk. The pathway is designed to prioritize early identification of this low-risk group of patients who have a low rate of MI at presentation or subsequent events over the next 30 days. We think that these patients can therefore safely be discharged with just a single test. It's also designed to reduce the number of patients who might need serial testing within the emergency department itself. And so its aim, the aim is to reach an admission or discharge decision in the majority of patients after just a single test. And we think that will increase the throughput of these patients through the emergency department.

And this pathway has been evaluated in a stepped-wedge cluster randomized trial known as the HiSTORIC trial. And we found that 20% more patients could be discharged directly from the ED, and the mean length of stay in the ED fell by three hours when this trial was implemented. And so, as Luca mentioned, for the current high-sensitivity assay, the threshold we use to identify low-risk patients we selected is five nanograms per liter. So on the current assay, that was the value below which troponin could not be detected by that assay. What is felt to be a, you know, a sensible approach. Our hypothesis is that, you know, this assay's got better and we can now identify more patients with very low concentration values with more confidence in the accuracy of the test at that range. And our hypothesis is therefore, that with this assay, we might be able to raise that threshold and identify more patients who would be suitable for discharge early on in that journey.

And so, in the project we've completed, we were looking at how we might adapt the High-STEACS pathway for use with the new sixth generation high-sensitivity assay. So our aims were to derive an outcome-driven threshold to identify the group of low-risk patients. We wanted to evaluate the performance of the pathway in comparison to the current assay, with the current thresholds that have been validated and implemented in clinical practice, and then working with Luca in the group in Basel, we wanted to validate the performance of the pathway in an external cohort to confirm our findings.

So we conducted a pre-specified analysis of a study known as the POC-ET study. And this is adults presenting with possible MI to three different emergency departments across Scotland, in which blood samples were taken at zero, one, two and 6 to 36 hours. And similarly we had a process where each case was adjudicated by two separate physicians, with a third resolving any disagreements, in accordance with the Fourth Universal Definition of MI. Our primary outcome was Type 1, 4b, or 4c MI at presentation, or MI or cardiac death over the subsequent 30 days. Cardiac troponin was measured with the current generation of the assay and the novel sixth generation high-sensitivity assay.

And so for that first aim, about the selection of the low-risk stratification threshold, we looked for the highest value of troponin T Gen 6 that gave us a sensitivity of more than 99% for the primary outcome, and a negative predictive value of more than 99.5%. So no more than 1 in 200 patients of our low-risk group would go on to have either MI at that index presentation or events over the next 30 days. We calculated the diagnostic metrics for this threshold, the performance of our pathway, and then validated our findings in the APACE cohort, which Luca described to you.

So we recruited a thousand participants to POC-ET, of whom we had enough sample left over in 987 to measure the sixth generation assay, and then 923 of these patients had enough sample in two separate samples for us to analyze pathway performance. This is a predominantly male, middle-aged cohort, with a moderate prevalence of co-morbidities and cardiovascular risk factors. And 82 of our patients had MI at index presentation, one patient had an event, an MI over the next 30 days, two of the patients with index MI had cardiac death within 30 days.

So this slide just shows plots showing the concentrations on the fifth and the sixth generation assays. I think as you can see, the two axes are really tightly correlated. But it's very important to note that the calibration for the two assays has evolved since the fifth generation assay was released. And so values on the sixth generation assay tend to read a higher value off the same sample as the ones in the fifth generation assay. This means we cannot simply transport the thresholds we've previously used for the current assay onto the novel assay.

So onto selection of the low-risk threshold. So the bar plot in the middle shows the cumulative proportion of patients who were below a series of thresholds from very low up to the upper reference limit. And the top plot shows in red the negative predictive value for the primary outcome, and in blue the sensitivity for the primary outcome for each of those thresholds. And so what we found is a threshold of less than 13 nanograms per liter at presentation was the highest threshold identified. The most patients with a negative predictive value of more than 99.5%, and a sensitivity of more than 99%.

In comparison to the fifth generation threshold, this identified a lot more patients. We discussed that the threshold that has been validated for the fifth generation assay was less than 5 ng/L, and we saw that more than double the patients at presentation could be ruled low risk here with very similar negative predictive value and sensitivity. We worked with our friends in Basel, in the APACE study and saw a very similar picture, in that the number of patients below the sixth generation threshold was considerably higher than the patients below their fifth generation threshold. And again, negative predictive value and sensitivity were very high.

So, just to show how we incorporated this into the pathway for analysis. So the low-risk stratification threshold becomes less than 13 nanograms per liter, as long as you've had more than two hours since the onset of symptoms. And you have no signs of ischemia on the ECG. High risk patients, anyone with a presentation troponin above the specific 99th percentile, that is 18 nanograms for women, 32 nanograms for men. And for those patients in the intermediate, a second test, if it doesn't change, is low risk. If it does change, well, it's high risk.

These flow diagrams just show how our patients were treated by the application of the pathway and at which step they were ruled either high risk or low risk. So those patients who are high risk in red, the low risk ones in blue. Those who met the primary outcome are shaded dark, whereas those patients who didn't meet the primary outcomes, had no MI, no cardiac death, shaded light. And you see, the first step, those with ischemia could become high risk. The second step, those with a low troponin at presentation, no ischemia on the ECG, become low risk. Those patients with a troponin over the reference limit become at high risk. And then for the intermediate group, a second test gets the definitive triage. And at the top we've got the sixth generation assay, in the bottom we've got the fifth generation assay.

What we see is that the differences between the two groups overall were quite similar. So slightly more patients were at low risk by the novel assay with a very similar negative predictive value and sensitivity to this approach, which has been validated and recommended in guidelines. However, the really significant difference we observed was that significantly more patients, an increase of 18 to 41%, could be identified as low risk at presentation, so didn't need serial testing. The numbers that need serial testing fell from 44% to 23% of the cohort.

Obviously, it's important in any evaluation of a pathway to go and look at how your pathway performs in an external cohort. And this is the data we found for the APACE cohort. More patients in the cohort had MI. And you see therefore that fewer patients were low risk. But once again, more patients were ruled low risk after one test rather than two. And similarly, only about 1 in 4 patients required serial testing to definitively establish the risk. We saw significant overlap in the 95% confidence estimates, around the value for sensitivity. Given this is a slightly higher prevalence group, the NPV is slightly lower at that sensitivity.

So, the key messages from our study are that using an optimized risk stratification threshold with the novel sixth generation assay, we can potentially significantly increase the proportion of patients who could be identified as low risk after only a single test. We found similar sensitivity and NPV for myocardial infarction and cardiac death at 30 days compared to the strategy using the fifth generation assay, which has been validated and recommended in the guidelines.

Of course, this is observational data, and we need prospective studies where the sixth generation assay is used to guide care, to really demonstrate safety and evaluate whether we see a reduction in serial testing in real life and whether that translates to a reduced length of stay in the emergency department. We think that could really increase the throughput for these patients and hopefully help tackle the issue of crowding, at least from the angle we can help with patients with chest pain.

Of course, it's important to highlight a few limitations. So only 30% of our patients were female, and we'd have been really interested in evaluating a sex-specific single rule-out threshold, because the separation between male and female patients with this assay is more marked at very low concentrations than we've perhaps seen with previous assays. A thousand patients is a comparatively small sample size in the context of the literature that's available for previous assays, where thresholds have been validated in meta-analyses of more than 10,000 patients. We hope that as more data for the sixth generation assay becomes available, we'll be able to conduct a similar kind of meta-analysis or approach to increase confidence. So, thank you very much for your attention. And I'd like to thank all of my collaborators, who played an invaluable role. Obviously, nobody can do any of these studies alone. And particularly I'd like to thank the patients who participated.