BACKGROUND: We undertook an assessment of current use of evidence-based guidelines for the use of cardiac biomarkers in Europe (EU) and North America (NA).
METHODS: In 2013–2014 a web-based questionnaire was distributed via NA and EU biochemical societies. Questions covered cardiac biomarkers measured, analytical methods used, decision thresholds, and use of decision-making protocols. Results were collated using a central database and analyzed using comparative and descriptive nonparametric statistics.
RESULTS: In EU, returns were obtained from 442 hospitals, 50% central or university hospitals, and 39% from local hospitals from 35 countries with 395/442 (89%) provided an acute service. In NA there were 91 responses (63.7% central or university hospitals, 19.8% community hospitals) with 76/91 (83.5%) providing an acute service. Cardiac troponin was the preferred cardiac biomarker in 99.5% (EU) and 98.7% (NA), and the first line marker in 97.7% (EU) and 97.4% (NA). There were important differences in the choice of decision limits and their derivations. The origin of the information was also significantly different, with EU vs NA as follows: package insert, 61.9% vs 40%; publications, 17.1% vs 15.0%; local clinical or analytical validation choice, 21.0% vs 45.0%; P = 0.0003.
CONCLUSIONS: There are significant differences between EU and NA use of cardiac biomarkers. This probably relates to different availability of assays between EU and NA (such as high-sensitivity troponin assays) and different laboratory practices on assay introduction (greater local evaluation of assay performance occurred in NA).
Measurement of cardiac troponin has become accepted as the preferred biomarker of myocardial injury (1), endorsed by the proposed redefinition of myocardial infarction (MI)14 in the year 2000 (2), and incorporated into management guidelines produced by the European Society of Cardiology (ESC) (3) and the American Heart Association (4, 5). The universal definition of MI recommends the use of the 99th percentile as the diagnostic discriminant (6, 7). Further, high-sensitivity (hs) assays have been developed for cardiac troponin T (cTnT) (8–10) and cTnI (11, 12). The value of hs-cTnT and cTnI assays for rapid MI diagnosis has been recognized and is now part of recommended guidance (13).
The Working Group on Cardiac Markers of the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) has undertaken a series of audits to estimate the uptake and utilization of the recommendations from expert bodies in the clinical and laboratory community on the use of cardiac biomarkers in patients with suspected MI. In addition, this group recently undertook a small pilot study of North American laboratories to compare practice. The aim of this paper is to document the adherence to clinical practice guideline recommendations for cardiac biomarker use in Europe and North America and to give a measure of any changes or improvement in adherence over the last 7 years (14–16).
The CARMAGUE (CARdiac MArker Uptake of Guidelines in Europe) study was a questionnaire-based audit of current laboratory practice compared with international recommendations and guidelines. The questionnaire was compiled with experience from the 2 previous questionnaires and also incorporated some modifications in design to allow more in-depth analysis of selected responses. The questionnaire included the biomarker repertoire offered by the participating laboratories and the methods used. It was particularly concerned with the adherence to the current recommendations for diagnosis of acute MI (use of the 99th percentile threshold) and the evidence used by the participating laboratories for the selection of their diagnostic discriminant. The questions were sequentially structured using multiple choice answers to distinguish the diagnostic discriminant employed from the source of the value selected. The degree of discussion between the laboratory and clinician communities on diagnostic pathways and collaborative working was also investigated. The audit also covered general laboratory practice such as accreditation status and quality assurance practices. Invitation to participate was distributed as a letter containing a web link via participating national societies of the EFLM and for North America by the AACC. The final questionnaire used in the survey is displayed at http://www.carmague.fi/1/. The survey was first distributed in early 2013 and completed at the end of that year.
In the European survey, a total of 493 responses were received from 36 countries. Of these, 442 (89.6%) were sufficiently complete to allow inclusion in the analysis. Fifty percent of the responses were from central or teaching hospitals and 39% from district or community hospitals. This distribution was not statistically significantly different from the 2 previous studies (14–16) (respectively for central and district hospitals 55% and 39%, 58% and 35%).
Because the questionnaire was distributed as an electronic link rather than posted, response rate could only be estimated from the reported number of hospitals or hospital groups delivering acute services in each country, where available (n = 31), as well as the absolute numbers of returns. This is likely to underestimate rather than overestimate response rate. For example, in England, 500 hospitals are listed but there are only 154 acute service providers (owing to formation of multihospital groups). In addition, regionalization of laboratories and hospital networks would be expected to standardize interpretive protocols. The majority of returns, like the previous surveys, were from northern European countries. Estimated return rates were as follows: 6/36, ≥50%; 8/36, 20%–49.9%; 2/36, 10%–19.9%; 5/36, 5%–9.9%; and the remainder (15), <5%. Southern European countries were noticeably underrepresented with very few returns from Italy, Spain, Portugal, and France. Although there were a reasonable number of returns from Germany (50), this was estimated at only 6.1% of hospitals but 64% were university or regional hospitals.
In this round of the survey, a significant proportion (11%) of those who responded did not provide a 24-h service (compared to 2%–5% on the previous surveys). Some of the data comparisons of the current results with those of the previous surveys therefore excluded laboratories from the present survey where a 24-h service was not provided. When such an analysis has occurred, it is indicated in the text. The North American survey returned a total of 135 responses. This included nonclinical institutions and some hospitals that were not located in North America (a result of the international nature of AACC membership). Ninety-one responses containing useable data were obtained from North American laboratories that provided clinical patient services. The distribution of laboratory types was different between the 2 areas with more laboratories (16.5%) in North America than in Europe not providing a 24-h service. Of the 76 laboratories that provided a 24-h service, 58 (63.7%) were university or regional hospitals and only 18 (19.8%) were district or community hospitals. There were significantly more university and central hospitals in the North American sample than in the European sample (P = 0.0011).
UTILIZATION OF TROPONIN
Cardiac troponin was the most commonly offered biomarker test for diagnosis of MI, offered in 98.6% of all European and in 91.2% of all North American laboratories. In laboratories offering a 24-h service, 99.5% of European and 98.7% of North American laboratories similarly offered cardiac troponin for diagnosis of MI. In the European laboratories, sample cost was the only reason cited in the 6 laboratories that did not offer cardiac troponin in the cardiac biomarker repertoire. For North America, cost plus clinician preference was cited for the 1 laboratory not offering cardiac troponin that provided a 24-h service. Overall, the number of hospitals offering has increased slightly from 2006 to 2013 (95.3% to 98.6%) but this change was not statistically significant. Cardiac troponin was offered on its own as the first line test in 35.8% of European laboratories offering a 24-h service. The overall trend for use of cardiac troponin as sole biomarker was for an increase in utilization (28.6% in 2006, 31.4% in 2010) but this failed to reach statistical significance. In contrast 50.7% of North American laboratories offering a 24-h service used cardiac troponin as the only biomarker (P = 0.0223 vs European laboratories).
OTHER CARDIAC BIOMARKERS
From the original survey (14–16), there has been a significant shift in the use of biomarker measurements. In accordance with recommendations, there has been a significant reduction in the use of biomarkers, which are no longer recommended (Fig. 1) P <0.0001 for aspartate transaminase (AST), lactate dehydrogenase (LDH) or hydroxybutyrate dehydrogenase (HBD) and creatine kinase (CK) but a retention of creatine kinase MB isoenzyme (CK-MB) activity measurement. Comparing the cardiac marker profiles currently offered in Europe and North America by laboratories delivering a 24-h service (Fig. 2) there are significant differences. Both North American and European laboratories offer CK-MB mass and troponin but in addition European laboratories offer CK (P < 0.0001) and CK-MB activity (P = 0.0009). Both in Europe and in North America laboratories reported retaining lactate dehydrogenase/hydroxybutyric dehydrogenase (LD/HBD) on their cardiac test repertoire but significantly more in Europe than in North America (P = 0.0017). Clinician familiarity was the reason most often cited for retention of other markers in both Europe and North America; in Europe, for CK the rate of citing this reason was 120/165 (73%) laboratories, for CK-MB activity it was 72/97 (74%), and for LD/HBD it was 42/69 (61%). For North America this reason was cited for CK in 9/11 (82%) laboratories, CK-MB activity in 4/6 (67%), and LD in 1/2 (50%).
There has been a significant change in the use of the evidence base for decision making (Fig. 3) with a shift to use of the 99th percentile (P < 0.0001) from 2006 to 2013. There is a significant difference between current European and North American practice (P = 0.0197). More European laboratories use the 99th percentile or locally decided cut points whereas more North American laboratories utilize assay imprecision (Table 1). In Europe, when assay imprecision or the 99th percentile was not used, local clinical decision was the largest single category, 45/102 (44%) although 27/102 (26%) laboratories reported obtaining information from the package insert. In North America 11/14 (79%) of those who did not use analytical imprecision or the 99th percentile used a local clinical decision.
Derivation of the decision limit has also shifted significantly (Table 2) for European laboratories with more laboratories now using the laboratory package insert when comparing data from 2006 and 2010 with current practice (P <0001). North American laboratories are more likely to undertake local validation of the manufacturers' claims (Table 2). It is noteworthy that in Europe, hs-cTnT is available and that this assay is now used by 208/221 (94.1%) of laboratories measuring cTnT compared to 2/124 (1.6%) in the last survey. Only 2 European laboratories were using hs-cTnI in the current survey. In contrast, hs assays are not available in the US.
A detailed analysis of the decision limits used was performed for the 2 most commonly used troponin methods, the Roche hs-cTnT and the contemporary Abbott cTnI assays. The range of cutoff used for hs-cTnT was 2 to 700 ng/L with 50.3% using the 99th percentile value of 14 ng/L. The other values most frequently used were 30 ng/L, 50 ng/L, and 100 ng/L. In the subgroup of laboratories that claimed to use the 99th percentile, only 76.8% used 14 ng/L. A detailed distribution of this data is shown in Table 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol62/issue9. Comparing decision limits between the hs and conventional versions of the cardiac troponin assay showed that the majority of the conventional assay users selected a discriminant of 50 or 100 ng/L (see online Supplemental Table 1). Similarly, although numbers are small, in Canada where a hs troponin T is available 5/7 respondents used a discriminant of 14 ng/L (although 2 were still using higher values) whereas in the US, where the fourth-generation cTnT assay is in use, 4/10 were using a discriminant of 10 ng/L and 4/10 using a discriminant of 100 ng/L. Data for the Abbott cTnI assay were similar, with a range from 25 to 500 ng/L and peaks of utilization at 30 ng/L, 40 ng/L, and 300 ng/L. A detailed analysis of the results is shown in online Supplemental Table 2. Only 12.1% of those claiming to use the 99th percentile used the appropriate value of 28 ng/L and only 15.4% of those using the 10% CV similarly used the appropriate value of 30 ng/L. A similar pattern was seen for all of the assays examined both in the European and the North American study, with no single value being used consistently and where there were sufficient data for a more detailed analysis, no consistent use of either the correct 99th percentile or 10% CV value. These data are summarized in online Supplemental Table 3.
The Universal Definition of MI requires repeat (serial) testing for diagnosis. The sampling intervals are method dependent, with 3 h sampling recommended for hs assays but 6-h sampling recommended for assays that are not hs. In the European sample (342/393 responses), 240/342 (70.2%) used serial testing in all cases. Of the 240 l used serial testing, 64.2% did not use an absolute or relative δ, 7.9% an absolute δ, 21.7% a relative δ, and 6.3% used both. In the North American sample with 61/76 responses, 46/61 (75.4%) used serial testing in all cases with no statistically significant difference between the 2 geographical areas (P = 0.59); 81.9% did not use an absolute or relative δ, 5.7% an absolute, 1.9% a relative δ, and 7.5% both. More laboratories in the European sample used a δ change than in the North American sample (P = 0.02).
Comparison of sampling intervals is shown in Table 3. For both regions a 3-h or 6-h interval between consecutive samples was the most common sampling protocol with no difference in frequency between European and North American laboratories when laboratories using a high-sensitivity assay were excluded. In Europe, comparison of laboratories using a high-sensitivity assay with those using a contemporary sensitive assay showed that if a high-sensitivity assay was used, a 3-h sampling interval was used more frequently than a 6-h sampling interval (P = 0.0097).
There has been a progressive and significant increase in laboratories reporting a protocol that was an agreed protocol with their clinical colleagues from 49/220 (22.3%) in 2006 to 148/303 (48.8%) in 2010 and 267/395 (67.6%) in the current survey (P < 0.0001). In North America hospitals providing acute services, 50/76 (65.8%) also reported an agreed protocol. However, it was surprising to note that only a minority of laboratories in both Europe (35.2%) and in North America (36%) reported the use of a written agreed protocol. The majority reported either an informal consensus or a verbal agreement with the clinicians (see online Supplemental Table 4). There was no difference in the reported frequency of arrangements between Europe and North America.
The principal results of the survey can be summarized as follows. First, since the introduction of the redefinition and then the Universal Definition of MI, troponin has been rapidly adopted as the principal biomarker of acute myocardial injury and is tending to become the sole marker. Second, the use of other biomarkers has variably declined, with biomarkers known to show poor specificity being abandoned. However CK, CK-MB, and LDH testing appears to be retained. Third, there has been a shift toward the use of the 99th percentile as the clinical discrimination limit. Finally, and reassuringly, there is also an increased interaction between laboratory professionals and clinicians. This is also enforced by current medical laboratory requirements for quality and competence, such as EN ISO 15189:2012. It is likely that the final outcome would be the greater use of the 99th percentile and the cessation of use of outdated tests. Although findings in Europe were similar to those in North America, there was a difference regarding implementation of the 99th percentile as a diagnostic cutoff for MI. This may be driven by the lack of availability (not cleared for use by the US Food and Drug Administration) of high-sensitivity assays in the US, whereas they are available in Canada. However, although a lower cutoff can be selected in the responses for troponin T even with the fourth-generation assay, this was only seen in 40% of the respondents in the US.
Although these findings indicate an encouraging trend, a more detailed analysis of the data reveals a significant lack of application of existing evidence-based guidelines. It is now 15 years since the redefinition of acute MI with cardiac troponin designated as the gold standard biomarker. Early studies demonstrated the superiority of cardiac troponin to CK-MB and CK (17–19). The Universal Definition of MI was subsequently published in 2007 as a worldwide standard (6). The near universal adoption of cardiac troponin as the cardiac biomarker of choice has occurred with the only imperative for not measuring cardiac troponin being cost. However, the cost of cardiac troponin measurement has fallen dramatically and is now comparable to that of measurement of CK-MB mass measurement. As cardiac troponin measurement can be incorporated into clinical decision-making pathways that allow efficient resource utilization (20, 21) the argument for not measuring cardiac troponin on cost grounds cannot be justified on clinical or cost effectiveness criteria (22).
The retention of LDH may represent different clinical practice or some confusion in providing data for the questionnaire (although the questionnaire explicitly asked for markers included in the cardiac profile). Respondents indicated that the most common reason cited for retention was clinician familiarity. There is no clinical justification for continuing to use LDH as part of a cardiac profile as the half-life of cardiac troponin allows late diagnosis (23) and can be used for infarct sizing (24, 25). The redefinition of MI recommended that LD measurement should no longer be used in diagnosis (6). CK, CK-MB, and myoglobin continue to be measured. The most common justification of the use of these biomarkers found in the study was clinician familiarity, for early diagnosis, or as a reinfarction marker. The measurement of both contemporary sensitive and high-sensitivity troponin is superior to all of these markers for the early detection of acute myocardial injury (26–28). Similarly, it has been shown that serial troponin measurement is equivalent to measurement of CK-MB for detection of reinfarction (29). Measurement of CK-MB by activity measurement is clinically and methodologically inadequate (30, 31) and is not recommended (2) but was cited as performed on cost grounds and clinician familiarity. In sum, there is no longer justification for CK-MB activity or mass measurement.
The variability in the use of diagnostic discriminant suggests that the guidelines are not being followed. The consequence of this will be that the ability to apply data- or guidelines-suggested approaches cannot be implemented effectively. If interventions are based on a specific diagnostic pathway including a troponin measurement that classifies patients into different management strategies, there are potential clinical consequences. It has been very clearly demonstrated that the use of a diagnostic discriminant as close to the 99th percentile as can be achieved has significant clinical benefits in improved diagnostic accuracy and mortality reduction (32–34). The 10% CV, which is used as an alternative is not required if the 99th percentile can be measured with an imprecision of <20% (35). Although there has been a progressive shift with increasing use of the 99th percentile as the decision limit, it is apparent that many laboratories use a different diagnostic discriminant and, more worryingly, amongst those claiming to use the 99th percentile, a substantial number use a value, which is not in accordance with the information from the manufacturer or derived from published literature.
In addition, failure to use a consistent diagnostic discriminant for cardiac troponin will alter the number of patients classified as having had an MI. This will increase the heterogeneity for the diagnosis of MI in a way that will confound both clinicians and affect publications.
Finally, it should be noted that the degree of local validation of assay performance appears to be somewhat limited. As there is noteworthy dependence on manufacturer-provided information, it is extremely important that independent validation of the manufacturer's performance claims is performed by the laboratory community. Many laboratory medicine journals (with some notable exceptions) no longer publish papers that address the clinical and analytical performance of laboratory methods, although the recent advent of more methodologically focused journals may help to fill this gap.
The choice of sampling times used by the majority of laboratories is in accordance with the recommendations from biochemical and cardiological societies. The data for contemporary assays showed similar utilization patterns in Europe and in the US. A shift towards a 3-h sampling interval in Europe probably reflects the availability of high-sensitivity assays compared to the lack of access in the US to state-of-the-art assays. Laboratories using high-sensitivity assays have shifted towards a more aggressive strategy of measurement at 3-h intervals. This approach forms part of the previous two guideline recommendations of the ESC (13, 36) and has recently been endorsed in the UK in an evidence-based review by NICE (National Institute for Health and Care Excellence) (22). A rather surprising finding was the lack of a clearly defined protocol for assessment of patients presenting with chest pain. The laboratory community makes extensive use of standard operating procedures and well-defined protocols for all aspects of laboratory work, but only a minority of laboratories had a defined written agreement for the use of cardiac biomarkers within the hospital. This lack of direct clinical engagement was noted in the previous studies and does not seem to have improved appreciably. In the 10 commandments of troponin, clinician/laboratory dialogue is highlighted as the first commandment (37).
In conclusion, although there has been substantial progress in adoption of cardiac troponin measurement as the cardiac biomarker of choice for detection of acute myocardial injury and myocardial infarction, there is an important gap between existing recommendations and clinical and laboratory application. The largest single area where there is room for improvement is the adoption of the appropriate 99th percentile as diagnostic discriminant. Even where there is use of a high-sensitivity cardiac troponin assays, the current study shows a gap between recommendation and application. A combined goal of professional societies, both biochemical and cardiological, must be to encourage the adoption of evidence-based decision guidelines (38). As more laboratories switch to high-sensitivity assays, manufacturers should encourage the adoption of the true 99th percentile. However, existing assays should be used to their best capability, especially in those countries where there are regulatory problems in using high-sensitivity assays, to ensure that patients are not disadvantaged further from lack of access to contemporary immunoassay technology.
↵14 Nonstandard abbreviations:
- myocardial infarction;
- European Society of Cardiology;
- cardiac troponin T;
- European Federation of Clinical Chemistry and Laboratory Medicine;
- aspartate transaminase;
- lactate dehydrogenase;
- hydroxybutyrate dehydrogenase;
- creatine kinase;
- creatine kinase MB isoenzyme;
- lactate dehydrogenase/hydroxybutyric dehydrogenase.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: F.S. Apple, Clinical Chemistry, AACC.
Consultant or Advisory Role: F.S. Apple, Philips Healthcare Incubator; S. Stankovic, Roche.
Stock Ownership: None declared.
Honoraria: F.S. Apple, Metanomics Healthcare, Abbott Diagnostics, Siemens Healthcare, Roche Diagnostic, Alere, Nano Mixe, and Trinity.
Other Remuneration: F.S. Apple, HyTest.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: No sponsor was declared.
- Received for publication April 28, 2016.
- Accepted for publication June 16, 2016.
- © 2016 American Association for Clinical Chemistry